Apple releases eight small AI language models aimed at on-device use
In the world of AI, what might be called “small language models” have been growing in popularity recently because they can be run on a
This comprehensive guide describes fundamental sociolinguistic criteria for the safe, fair and ethical development and implementation of automatic interpreting products using artificial intelligence (AIxAI), also called machine interpreting.
12:31:15 So hi everyone. I’m the senior analyst at CSA research who worked on the perception survey for the Safe AI task force.
12:31:26 The interpreting industry is facing a tremendous amount of angst over improvements in artificial intelligence. Does it mean the end of the profession?
12:31:36 Will meaningful language access and meaningful communication access be maintained when using automated solutions, how can you ensure that AI is used safely to prevent negative outcomes?
12:31:50 There are a lot of questions to address. So, we conducted an in-depth research report and we recently published 350 pages packed with lots of information.
12:32:02 And this session I’m going to give you a summary of core data points that you will see analyzed in greater detail in the report.
12:32:11 But let’s start by discussing the context for this research. It was commissioned by the Safe AI task force.
12:32:19 It’s a diverse group of industry stakeholders who got together to establish disseminate and promote guidelines for responsible AI adoption in the field of interpreting.
12:32:30 And they requested that CSA research conducts for them a large-scale perception study of end users.
12:32:39 Here think doctors, patients, etc. And then of requesters those are the buyers of language services and then the providers of interpreting services and technology so that’s of course the language service providers, the interpreters, and the AI tech vendors.
12:32:54 The goal of the study was to capture current perceptions about both spoken and signed AI for interpreting with a focus on the US market.
12:33:06 And for those of you who don’t know us, CSA research is an independent market research company.
12:33:11 We’ve been around for 20 years. And we focus exclusively on the language services industry. Whether the supply side or the demand side.
12:33:20 And the scope of the research that the task force requested was massive. So, we are grateful to the sponsors that the task force recruited to help fund this research.
12:33:31 Their names are on screen here and this research would not have been possible if not for these companies backing the project.
12:33:39 And I also want to acknowledge our sign language interpretation right now with Master Word who is donating the sign language interpretation for the session today.
12:33:51 The report will synthesize the results of 2 surveys. One was with requesters and providers of language services.
12:34:00 So here again, think procurement teams, schedulers, interpreters, etc. And then the other was with the end users.
12:34:05 Again, things like doctor patient and all those permutations. That end user survey was available in 10 languages.
12:34:13 And a great promotion effort from the task force led to over 2,500 responses from 82 different countries.
12:34:21 In the United States, we were able to collect responses from 48 different states. So, it had a broad reach.
12:34:28 Altogether, we collected 118 data points which we correlated against each other and that allowed us to publish 9,400 values.
12:34:38 And we also offered ample opportunities for respondents to provide free text answers and we collected an amazing 3,400 comments and that told over 62,000 words.
12:34:51 Some people honestly wrote entire essays. So, we’ve read and analyzed every single one of them and we sprinkled anonymized approach throughout the report.
12:35:03 And I briefly want to touch on the profile respondents because it did affect the results. Two-thirds of respondents were either spoken or signed language interpreters.
12:35:13 We had targeted 11 different groups. So that means that the average is influenced greatly by interpreter responses.
12:35:20 And that’s why each graphic in the report will indicate responses by group so you can contrast responses from their different perspectives.
12:35:29 We also had 69% of respondents that had some sort of a connection to the healthcare field.
12:35:36 And this is important because it’s a sector where mistakes can have life and death consequences. So, the survey collected more negative reactions to AI than if more respondents had come from other domains that are less high risk.
12:35:51 And then only 11% of respondents across our sample had either moderate or extensive experience. With automated spoken or same language solutions.
12:36:03 And that honestly was a little bit of a problem because perceptions ended up being based on opinions more than on facts.
12:36:09 So on each graphic we distinguish responses from those who had experience and those that have either a little or not.
12:36:17 And I also want to mention that 79% of respondents came from the United States. That was the primary recruitment target for the task force.
12:36:28 On graphics, you will see respondents in the United States versus outside US. Outside the US tended to be a bit more open to AI.
12:36:37 But that could have had to do with the fact that the survey might have reached predominantly people who were already interested in the topic…..
12:36:44 …in those other geographies.
12:36:47 Okay, so let’s get started with context data. This is from end users. Meaning either the service recipients or frontline professionals who participate in sessions where the interpreting occurs.
12:37:00 We ask them to what degree they trust in interpreting from 5 different sources. And what you see here, the results for those who say they either, that they fully trust the source.
12:37:11 So. No surprise, face to face interpreters come in the lead with 78% of end users who fully trust them.
12:37:20 Once the interpreter is remote, that confidence level drops to just 56%. Bilingual friends and family members who are not professionals did not rank very high here with only 21% of end users who fully trust them.
12:37:35 The last 2 options refer to the automated interpreting options. So, when an organization provides interpreting, think for example about a conference where the organizer gives you a link to access the automated interpreting.
12:37:51 In such cases respondents trust the output a little bit more than when using apps that they found online, often those free apps that you can just download by yourself.
12:38:03 But either way, it’s not very much trust in either of these 2 cases.
12:38:09 Let’s now explore a series of 5 AI powered services that’s really to the topic automated captioning.
12:38:17 Automated transcription, automated subtitling, automated spoken language interpreting, and automated sign language interpreting.
12:38:26 And we asked survey takers about their level of experience with it and if they had used it, what they thought about the quality.
12:38:34 Note that the end users of interpreting services did not see captioning, transcription, and subtitling questions because we were trying to simplify the survey for them.
12:38:47 So let’s start with captions. That’s when the text on screen appears in the same language as the one spoken in the session over a quarter of respondents had either used an automated system to produce captions or tested either moderately or extensively.
12:39:03 That’s actually a pretty good number. And over half of the M found results either good or excellent.
12:39:12 And this was the highest-level performance across the 5 services that we tracked.
12:39:17 The next service is related to captioning. But instead of the text appearing on a screen, the AI produces a document that may include elements such as time codes and who is speaking.
12:39:29 A little fewer people had experience with automated transcription. But overall, over half again found results either good or excellence.
12:39:38 What such high numbers indicate is the potential for technology to be useful to assist in some situations. Moving on to subtitling, the third best performing AI service.
12:39:51 That’s when captions on screen in a different language than the language spoken in the session. 19% of respondents had some experience with automation and 40% of them found some value in the results.
12:40:05 Numbers are a little bit less exciting than the captions because you compound errors at each step you add in the process.
12:40:12 So to simplify the concept in captioning, you essentially just use voice recognition. I’m simplifying here.
12:40:20 But for subtitling, you apply machine translation to the captions. So, you have doubled the chance of errors in the process.
12:40:29 Now, once you apply a synthetic voice to the subtitles, it’s called automated interpreting, machine interpreting, automated speech translation, the terminology out there is a necessarily very formalized what we predominantly use in the report is automated in interpreting or AI.
12:40:49 And that is the focus of the report. Unfortunately, only 11% of respondents had first-hand experience with such technology and just a little over a third of respondents found results either good or excellent.
12:41:05 One of the challenges of automated interpreting is that very often people prefer to read subtitles than to listen to a robotic sounding voice that lacks a little bit in intonation.
12:41:19 And finally, the last AI driven service we asked about was automated signed language, which refers to text to sign technology using avatars.
12:41:27 Assigned to text technology using complex sign decoding systems. And the reality is that the technology is much less developed than spoken language interpreting.
12:41:39 So barely 1% of respondents had even experienced any of it. And for those who did, quality perceptionss fell below the 30% ….
12:41:46 … mark, meaning that the use cases are going to be very limited. Until the technology improves.
12:41:53 Now I also want to show you some data on whether respondents thought that automated interpreting could reach the same level of accuracy as qualified human interpreters.
12:42:04 And for simple conversations 9%. Believe that AI already reaches human parity. And 25% believe it will happen soon.
12:42:14 Numbers naturally drop for complex conversations with a single percentage point who believe human parity already exists and 8% who think it will happen soon.
12:42:25 The flip side of course is that a large portion of respondents think it will take a really long time to get there.
12:42:32 It may never happen at all. And that’s why it gets very interesting to see the response difference between those who already have some experience and those who don’t.
12:42:42 What’s in the report is that respondents without much or any AI experience tend to underestimate AIs capability.
12:42:53 So let me prove this to you. We assigned a score. To quality perceptions, those who say quality is excellent, they have a value of 2, good is one, unacceptable, sorry, then I have 4, that’s minus one, an acceptable minus 2.
12:43:09 And that’s what you see on the x, the y-axis here, and then on the x-axis you see the level of experience.
12:43:17 And for each of the 5 AI driven services that we tracked, the more experience people have, the more positively they think about AI capabilities.
12:43:28 The effect on the results is very real. So, I highly encourage providers to gain some experience with it, interpreters in particular had the least experience with AI.
12:43:41 You should really test some tools so that you know what it can or cannot do. Otherwise, you only say what it should or should not do.
12:43:50 That’s important too, but that is not all there is to this discussion.
12:43:57 Okay, moving on I want to know show you a series of 10 statements that we made where we asked survey respondents to tell us to what degree they agreed with them.
12:44:11 And I’m going to show you responses for those who agree. The first statement was, it’s okay to use machines for routine and repetitive conversations that they can handle.
12:44:21 Nearly half of respondents either agree or strongly agree. And this confirms what we saw earlier about simple conversations.
12:44:30 AI can be helpful for some easier interactions or one-way communications. The next statement inquired whether it’s better to have a machine interpret rather than have no interpretation at all.
12:44:43 And this is the whole premise behind the argument that AI increases language and communication access. 44% of respondents either agreed or strongly agreed.
12:44:53 That imperfect, results are better than no language support. Now that leaves a lot of people who disagree.
12:45:01 Not also that respondents commented on the fact that AI could actually reduce meaningful language and communication access. Because organizations would push AI options in unsuitable scenarios.
12:45:15 If language and communication access decreases, or reduces in quality, then discrimination of physical harm or many other consequences can occur.
12:45:27 And obviously that is not something that anybody wants to see happen.
12:45:32 The next statement epitomizes the contradiction of the results throughout the reports. Many respondents claim that machines cannot deliver when they haven’t retested the systems.
12:45:45 Their answers are often more about saying what machines should be tasked with handling interpreting services.
12:45:54 And that is why nearly 3 in 4 respondents are whopping 74% either agree or strongly agree that it is not right to replace people with machines for interpreting.
12:46:06 Now remember we had a lot of responses from providers who are afraid to lose their job. So that affected the results to some degree.
12:46:15 And when you contract the responses of the previous ethics question to the potential financial motives of organizations to push for AI, you really get to the crux of the issue.
12:46:26 Over 2, of 3, of respondents either agree or strongly agree that requests is only want automation to reduce costs.
12:46:33 The ones who really disagreed with that statement were people in procurement role, which is normal because they see the whole picture with all the elements that go into such decisions.
12:46:45 Procurement teams try to fix problems for their companies and their primary motives are efficiency even before cost.
12:46:53 So, for example, a language professional may not see the cost of the administrative burden involved in scheduling interpreters.
12:47:01 Or what happens when there is a no-show from either the service recipient or the interpreter?
12:47:06 Does that mean the only answer to their problem is AI? Of course not, but you need to understand that buyers are frustrated with an inadequate scheduling system, and the lack of system integration and interpreter demands even….
12:47:20 …can affect decisions regarding the role of human versus AI.
12:47:26 Continuing the analysis and cost, we asked end users whether who pays for the interpreting effects, their thoughts on AI usage.
12:47:34 39% either agree or strongly agree that they might choose a machine if they have to pay for the interpreter themselves.
12:47:41 Note, of course, that the balance would still prefer a human interpreter. And this point is tied to the proliferation of pocket translators.
12:47:53 These apps you can find online for generally tourists and that affects the price point that technology vendors can charge.
12:47:59. And then this in turn drives the AI developers to create products primarily for enterprises and government institutions because developing quality applications is very costly.
12:48:14 And they need a return on their investment. And this also explains why some of the better AI products are not accessible to direct consumers.
12:48:25 Next statements, was one where we asked the opposite of the previous one. We’re contrasting payment by someone else versus by the end users themselves.
12:48:36 And if someone else pays the bill fully, 76% of end users will favor a human interpreter.
12:48:43 This shift in pay is important to remember in the context of respondents who feel that organizations push AI as a cost saving measure.
12:48:51 The cost factor matters for requesters and users alike. After all, this is interpreting. Nobody wants to pay for it.
12:48:59 Even though it is very necessary for all the people who are involved in communication.
12:49:06 Now switching gales a little bit the next statement was on whether there are situations where end users would prefer an automated interpreting solution over a human.
12:49:16 AI admittedly comes with its fair share of concerns over data privacy. However, being able to hold a conversation away from the ears and eyes of a fellow human might have some advantages too, especially in small communities where you personally know the interpreter.
12:49:34 Using a machine can be more comfortable when dealing with matters of a highly personal nature, anything that is sensitive or very private because despite the humanity that an interpreter brings to the table, sometimes that human element can be a little bit too much because here you can see that 31% agreed that sometimes AI could be better.
12:49:56 For them. And finally, the stress factor of using an automated solution also enters into the discussion. About half of respondents expect that using a machine interpreter would cause them no stress.
12:50:13 So why is that? AI driven processes add a burden to participants to figure out if a mistake was made and when it was made to Speak up if he did.
12:50:24 And they will likely fear loss of vital information that could affect a ruling, a diagnosis, or whatever the next step action is going to be.
12:50:33 Each of which can have a huge financial and human cost. And ultimately, distress can contribute to the feeling of not being heard.
12:50:45 And I also want to add another element is that the robotic nature of some of the AI tools output may be distracting and it might make it more difficult for the listeners to pay attention to the content and the next steps.
12:51:02 And we close a series of statements with a question for end users of whether they want to know if a person or a machine is doing the interpreting.
12:51:10 And the answer here was an overwhelming yes for 89% of respondents. Advances in advances in voice flow.
12:51:20 Can make the distinction between a human and a machine difficult when the end user doesn’t have a visual reference.
12:51:28 So you can use the analogy of phone systems which we all have struggled with it can become frustrating sometimes not to know if you’re talking to a human in our machine and the same would happen if you deal with a life translator versus a bot and not knowing which one you’re dealing with.
12:51:48 So now let’s tackle the pros and cons of automated solutions. I’ll present only the data portion of what’s in the reports note that there are dozens of pages in the report with an analysis of free text answers that went with it.
12:52:02 Respondents had a lot to say, especially when it comes to the drawbacks of potential A.
12:52:09 I use. But here I’ll strictly share the data portion. So, respondents could select multiple options from a list of common benefits.
12:52:17 Requesters and providers of interpreting services, so more options than end-user did.
12:52:23 You can see that with the little stars here in the legend. So, what does this graph tell us?
12:52:30 Well, the most frequent advantage was around the clock availability. 66% selected that advantage but end users and requesters were showing even greater numbers.
12:52:43 That’s when the interpreting solution is available, 24 7 365.
12:52:49 It’s technology that’s easy to implement. Now in second position, there was a tie between the no need to schedule.
12:52:56 Interpreter and the low cost element where 58% of respondents selected those responses. But what I also want you to pay attention to is the indirect dig against the interpreters.
12:53:10 Beyond the ability to skip the scheduling, some elements that respond positively about AI underscore issues even if they’re not significant.
12:53:18 But the issues that affect end users and requesters perception of human performance and therefore influence their reaction to AI.
12:53:28 So the numbers you see on screen at the average for all responses group. But let me show you examples from requests only meaning the buyers and decision makers.
12:53:38 38% like that AI doesn’t arrive late at appointments. 31%. Think that they could achieve higher field rates on job assignments.
12:53:46 And 16% think that they deal with fewer professionalism issues. So averages don’t tell the whole story, so pay attention when you read the report as to the difference between the different audiences, it’s on every graphic.
12:54:01 Okay, so now let’s examine responses for drawbacks. Big mistakes is where respondents worried the most.
12:54:10 We defined these big mistakes as situations when the main idea might be wrong or the mistake could cause harm or some sort of a legal problem.
12:54:20 And that’s where it’s 81% of respondents. When AI makes mistakes, the errors can be more severe than humans would make.
12:54:28 But humans make mistakes too, especially when they deliver services in real time. But in contrast though they can rebound a little better.
12:54:41 Now I also want you to look if you jump ahead a little bit here at the small mistakes. Those are situations where the main idea is clear, or the details are wrong.
12:54:55 That still worries 48%, but that number jumps to 57% for the end users.
12:55:01 So really compare the results by audience once you see the report because they tell different stories. Okay, so let’s go back to the beginning of the graphic here, the second bar.
12:55:11 The need for special tech worries another 67% of all respondents. The implementation of telephone and video.
12:55:22 Remote interpreting, launch the introduction of these dual handset phones and, you know, those physical carts that transport video devices.
12:55:30 And if you take the example of a medical setting, someone has to find a cart and maybe they didn’t order enough for their departments or maybe the last person who use it didn’t store it where it should have been and somebody has to hunt.
12:55:43 For it. So, if automated solutions use the same hardware, it will not really make the situation any easier for frontline professionals.
12:55:54 Okay, the third bar here respondents also fear something going wrong. 59% think there are more chances that something will go wrong with automated solutions.
12:56:03 And that always means more work for whoever must find a solution to troubleshoot and fix the issue. And then 56% fear that users will not accept machines.
12:56:15 Overcoming the stigma would require training and experience for all parties involved. But then on top of it for some end users, machines will never be the right solution.
12:56:26 Think for example about a patient who has multiple disabilities. There is a really high chance AI will not be able to assist them.
12:56:36 Okay, so most of the drawbacks are tied to the limitations of the AI and getting into the complex conversations where human interpreter is by far the superior option.
12:56:48 So that is why we ask survey takers if they thought that an automated interpreting system would be more useful if they could quickly get a person to help when there is a communication problem.
12:57:01 So. The logic behind the question is that the app would include some sort of a button where you can request to talk to a human interpreter.
12:57:09 Either by telephone, video, or even in person. You do that if either party in the conversation feels there’s a communication challenge.
12:57:19 It’s really similar to the press 0 to talk to a customer service representative in a phone-based system.
12:57:26 And some technology providers already built such an escalation mechanism into their platform. Here we found that more than one half of respondents think the ability to escalate to a human being would increase the usefulness of automated interpreting either a lot or level.
12:57:44 This ties a bit more to the symbiotic relationship. Between AI and interpreters.
12:57:51 So I’m not covering this here today, but in the report, we cover AI driven computer aided interpreting.
12:58:01 Which is when the AI helps with you know named entity recognition numbers etc. memory recall terms a lot of different aspects to empower the interpreter to do a better job.
12:58:17 I already mentioned that respondents cited a lot of concerns about AI performance and a lot of it relates to what machines cannot capture yet.
12:58:27 I’ll present a high-level categorization of what we explore more in depth in the report.
12:58:34 Respondents were truly prolific in their free text comments about the importance of humans in dealing with subtleties and ambiguities.
12:58:43 This graphic depicts a high-level categorization of their arguments from requiring the ability to deal with a context for language, culture, tone, emotions, and interaction background.
12:58:55 AI cannot capture visual clues, cultural inferences. It doesn’t understand the mood of the participants and it doesn’t really understand regionalisms.
12:59:05 So in short, the respondents fear that because AI cannot see nuance or read between the lines, service recipients have too much to lose when the machine doesn’t understand this context and culture.
12:59:18 But similarly, the language that people speak is not textbook perfect. Neither are the situations in which the interpreting occurs.
12:59:27 Respondents here were terrific too at providing examples of challenging situations where AI would struggle interpreting for someone who uses programmer, has a speech impediment or is in an active psychotic episode — that requires some serious skills.
12:59:43 So the graphic on screen shows some of the many such imperfect scenarios that can lead to a communication breakdown or technology failure.
12:59:53 The reason that humans remain the better choice is that they provide language accommodations for unusual or unexpected situations by adapting what and how they interpret their knowledge because they use the knowledge and their common sense.
13:00:09 And this takes us to an important concept that underpins the research. That of the role of the interpreter.
13:00:16 And I had a bit of fun with AI myself and I asked chat GPT to create a graphic depicting the traditional view of interpreting the one that revolves around the concept of linguistic conduit.
13:00:29 Meaning interpreters act as neutral parties who convey messages from one language to another without adding, omitting, or altering the content.
13:00:39 Ultimately, that’s what automated interpreting strives to accomplish. The thing though is that the interpreter’s role has evolved over time to be more than translators of spoken or signed words.
13:00:53 And in this active model, the interpreter plays a more involved role in the communication process. They may clarify meanings, ensure cultural appropriateness, and sometimes even mediate the conversation.
13:01:06 So in this model, the interpreters take responsibility to ensure successful communication. They work to verify the messages and pursue them correctly by both parties, potentially adapting language, tone, cultural references to fit the context.
13:01:22 So this approach recognizes the interpreter as an active participant in the communication process, acknowledging that their presence and their decisions can influence the outcome of the interaction.
13:01:35 AI is nowhere close to capable of reaching this level of proficiency. So, who will lose if AI pushes organizations to go back to a conduit model?
13:01:47 Of course, the end users. It may be less of an issue when interpreting for a business meeting or other, you know, trade show or conference but it is certainly in context such as healthcare, legal and social services.
13:02:03 And that then takes me to giving you a snapshot of data on AI suitability by use case scenario.
13:02:10 They are close to 6,000 data points and that alone in the report. So today I’m just giving you the most interesting ones.
13:02:18 Let’s examine end users’ responses first. The data you see here summarizes the percentage of service recipients and frontline professionals who find AI either mostly or totally suitable for a conversation type.
13:02:34 On the left you have low risk non-technical non urgent conversations that came up as the highest cause, while their logical counterparts on the right, high risk, technical and urgent, were seen as much less ready for deployments.
13:02:50 The trick of course is defining the characteristics of these conversation types as conversations rarely stick to a single type all the way through.
13:03:00 And if we now look at responses from requesters and providers, so no longer end users. Here you can see the top 5 use cases out of 58 that we provided across 11 areas.
13:03:14 Nearly 2 thirds of respondents of respondents think it’s mostly a totally suitable to use automated interpreting for notifications or announcements.
13:03:25 For example, when you report an outage, an absence, a cancellation, and I’ll skip ahead here and have you look at number 4 which has a similar situation but in an education context.
13:03:37 And then look at the type for number 5. That was an emergency service scenario. Where 37% believe that AI is okay to notify of a weather emergency.
13:03:50 But let’s go back to number 2 here, scheduling sessions to select, notify, or remind of either the time of the meeting where it will take place and what you need to prepare for it.
13:04:01 We found 56, sorry, 57%. Who found that AI was suitable for that.
13:04:09 And then in number 3, that came from our series on client service scenarios. That’s the ability to use AI.
13:04:16 I to check on the balance of an account. That’s something that is often frequently already done. Through AI phone systems as it is.
13:04:27 And then in number 5, the logistics call to explain how a session our conversation will take place, reaches 37% approval.
13:04:35 It’s a little similar to what was in number 2 here. Now these are very strong use cases.
13:04:41 Other sectors did not fare as well. But responses do vary significantly by use case, not just by domain.
13:04:50 So for example, if you take healthcare. The delivery of bad news to a patient is clearly unsuitable.
13:04:58 You all know this, and the data proved it. But if you’re doing patient registration, there was some potential there.
13:05:05 However, if you’re doing patient register, say with a patient that has cognitive differences, that would still be better with a human interpreter.
13:05:15 So there are a lot of elements that go into the decision. So that’s why the next part of the analysis identifies criteria to use when determining whether to use automated interpreting.
13:05:26 We provided respondents elements where they rated the criteria as to whether it should be a major criterion minor or not a criterion at all.
13:05:38 And I’m showing you here just the percentage for those who believe it’s a major factor. We told respondents to answer as if the decision was up to them.
13:05:48 The level of accuracy required from the interpreting leads the list, no surprise, followed by the risk of possible harm, no surprise there either.
13:05:58 And then comes the complexity of the language, not all language are rendered equally well through AI. But even after that you can see that the other elements, we suggest are all quite important.
13:06:11 And what you need to note is that one element alone is not enough to determine whether AI can or cannot do the job.
13:06:19 It’s always a multi-factor decision that is not just done at the organizational level, but for each single interaction.
13:06:28 So an organization might decide to procure an automated interpreting system. That doesn’t mean they should use it in every interaction.
13:06:37 Even if it’s a basic one. Okay, so I will wrap up this presentation of data highlights with responses from procurement teams about their plans for the next year for the 5 services that we had discussed upfront.
13:06:51 Note that the number of respondents here is very small. Not just 40 people who followed that survey path.
13:06:59 And the bulk of answers show no plans to use automated captioning, automated transcription, automated interpreting.
13:07:08 Subtitling or certain language, none of those. But let’s explore those who already went ahead with it or plan to do so.
13:07:15 So captioning and transcription are in the lead in terms of existing implementation. 18% of respondents have automated solutions.
13:07:25 In place for captioning. And 13% for transcription. Does that are 2 monolingual services that we’re tracked in the survey?
13:07:33 And only 5% which given the small sample and you know; of people this represents just 2 respondents.
13:07:43 Only 5% have solutions in place for interpreting and subtitling. None of them report any AI for sign language right now.
13:07:52 Now, there is little change planned on the radar. Those who were ready to implement automated captioning and sub-tiling already did so.
13:08:00 And implementation plans are negligible for other services. This thing is a bit different. All but sign languages will see some testing.
13:08:10 Truth be told, studies like the present one likely prompted some procurement teams to think it’s time to see what the fuss is all about.
13:08:19 And then what’s also interesting here in this data set is that more procurement teams plan to reduce their use of automated subtitling, than those who plan to continue using it.
13:08:31 And that is probably because the output is not up to expectations. That is problematic for the automated interpreting tech vendors.
13:08:42 Because they simplified mostly just add voice synthesis to the subtitles. Now we also notice some reductions for use of AI in captioning and transcription.
13:08:55 Okay, so I’m not going to take the time today to get into all the recommendations, which will be listed in the report for each of the audiences.
13:09:04 But I do want to give a flyby overview of the ones we gave for the task force. What’s important for everyone to remember?
13:09:10 Is that we just only scratch the surface, and we did not get anywhere close to enough feedback from end users.
13:09:17 The findings in this report alone are not sufficient. To establish all-encompassing guidelines. If you rush to establish guidelines, you will likely end up restricting use in situations where AI could have been a valid option.
13:09:31 So I want to urge you to remember your mission, which is to find out what is right for end users, not necessary to defend interpreters.
13:09:40 Or technology vendors. That was not the mission of the task force. So, what should you do? Now comes the hard work with more in-depth research to establish safe boundaries of when you can or cannot use AI.
13:09:56 So first you need studies with end users think focus groups and testing labs where end users can compare the different outputs.
13:10:05 Rate them, categorizing which scenarios they trust the system to do an adequate job. It should be for a variety of scenarios, not just the high-risk ones because realistically those are not ready for prime time.
13:10:18 The survey was also a little bit too theoretical for many end users who don’t even know what AI is or colony.
13:10:25 Imagine what automated outputs might sound like. Sweet, really needs to be hands-on. And you should compare different tools because the accuracy varies from one system to another.
13:10:37 It’s not equal out there. You should also conduct an in-depth analysis of the nature of conversation.
13:10:45 So we talked about simple versus complex conversations. And for each use case scenario you’re considering AI for, you need to really understand the percentage of conversations that are truly simple.
13:10:58 And when it turns complex. What are the trigger words that indicate the conversation tipped from simple to complex?
13:11:07 Because without that, we fall to any conversation having the chance to turn complex and then you like the data to put proper escalation systems when using AI.
13:11:19 And next our research barely touched on captioning and sub-titling. Automation use cases are much greater for those technologies and such research should identify the tipping point of when and users are satisfied with subtitling versus one date one voice synthesis instead.
13:11:36 And the same logic applied to captioning versus signed language by an avatar. Again, results will vary based on the scenario and the audience.
13:11:47 Finally, the presence perception study was conducted nearly too early to collect accurate perceptions in 6 months, a year or even 2 years, many more people will have been exposed to AI solutions and will have perceptions more in line with the capabilities.
13:12:05 Also, don’t forget technology is also constantly improving. So, what is not suitable now, would not be suitable in a matter of months.
13:12:15 So this is it for the presentation today. You can read the report online. On our platform or download a PDF, plan some time to read it.
13:12:25 It probably takes about 8 h to read it straight through. And that’s even at a pretty good pace.
13:12:29 It is packed with information, and I recommend you don’t skip the second chapter that tells you how to read the graphic because a graphic like the one you see on screen has about a hundred data points and you really need to understand.
13:12:42 What each of them represents to get the most out of the report. And I also want to mention that the 350 pages of the report are a bit daunting for some people, so there will be a summary version available pretty soon.
13:12:57 This concludes the presentation of the findings. Thank you everyone for watching. Thank you.
00:00:01:06 – 00:00:01:20
Hello.
00:00:01:20 – 00:00:03:23
Hello, everyone.
00:00:03:23 – 00:00:06:23
My name is Tim Riker
00:00:07:02 – 00:00:11:05
and I am going to be the presenter today,
00:00:11:12 – 00:00:12:11
one of the presenters.
00:00:12:11 – 00:00:13:22
I’m from Brown University
00:00:13:22 – 00:00:17:05
and I’m a member of this advisory board,
00:00:17:05 – 00:00:19:21
which is for artificial intelligence
00:00:19:21 – 00:00:22:03
and sign language interpreting.
00:00:22:03 – 00:00:23:19
Today we are thrilled
00:00:23:19 – 00:00:26:20
because we are going to be providing you
00:00:26:20 – 00:00:28:04
a presentation.
00:00:28:04 – 00:00:30:08
We’ll be talking about our report.
00:00:30:08 – 00:00:32:18
And
00:00:32:18 – 00:00:34:17
the advisory council is here today
00:00:34:17 – 00:00:35:20
together.
00:00:35:20 – 00:00:37:07
We’ll be talking about some of the work
00:00:37:07 – 00:00:39:08
that we’ve done, collecting data
00:00:39:08 – 00:00:40:17
from three webinars
00:00:40:17 – 00:00:43:09
that we hosted last fall.
00:00:43:09 – 00:00:45:01
The reason that we decided
00:00:45:01 – 00:00:46:05
to host these webinars
00:00:46:05 – 00:00:47:02
and do this research
00:00:47:02 – 00:00:48:17
is because we wanted to get more
00:00:48:17 – 00:00:50:06
of the deaf perspective
00:00:50:06 – 00:00:51:23
and to take that perspective.
00:00:51:23 – 00:00:54:00
And we do
00:00:54:00 – 00:00:55:10
some more research and former
00:00:55:10 – 00:00:57:20
Task Force for Safe AI.
00:00:57:20 – 00:01:00:09
That report is going to be presented
00:01:00:09 – 00:01:03:16
today and we
00:01:05:02 – 00:01:06:10
would like to share now with you
00:01:06:10 – 00:01:09:10
the topic of the session.
00:01:12:02 – 00:01:14:15
I’ll go ahead and introduce myself.
00:01:14:15 – 00:01:17:06
And so in terms of visual description,
00:01:17:06 – 00:01:20:11
I am currently wearing a black shirt.
00:01:21:11 – 00:01:22:07
It’s a long sleeve
00:01:22:07 – 00:01:23:04
black shirt,
00:01:23:04 – 00:01:24:09
collared shirt
00:01:24:09 – 00:01:27:09
with buttons and a white male
00:01:27:16 – 00:01:30:16
with reddish blondish hair.
00:01:30:21 – 00:01:33:20
I have a bit of a mustache
00:01:33:20 – 00:01:36:14
and facial hair.
00:01:36:14 – 00:01:38:23
So today our presentation
00:01:38:23 – 00:01:42:00
topic is going to be death safety.
00:01:42:07 – 00:01:43:23
I
00:01:43:23 – 00:01:47:03
a.i meaning artificial intelligence.
00:01:47:15 – 00:01:48:20
And our goal
00:01:48:20 – 00:01:50:12
is to have a legal foundation
00:01:50:12 – 00:01:53:12
for ubiquitous automatic interpreting,
00:01:53:17 – 00:01:55:19
using artificial intelligence
00:01:55:19 – 00:01:58:19
and how that relates to interpreting.
00:01:58:20 – 00:02:01:16
So
00:02:01:16 – 00:02:04:05
I wanted to talk a bit about why
00:02:04:05 – 00:02:07:05
this topic is so important.
00:02:08:24 – 00:02:09:21
As you know,
00:02:09:21 – 00:02:11:23
there are many users
00:02:11:23 – 00:02:13:13
of sign language interpreters.
00:02:13:13 – 00:02:17:17
We sometimes go through frustrations
00:02:18:02 – 00:02:19:03
and situations
00:02:19:03 – 00:02:22:04
because we are using technology,
00:02:22:13 – 00:02:25:07
and sometimes technology
00:02:25:07 – 00:02:26:10
can be very beneficial,
00:02:26:10 – 00:02:28:09
while other times technology
00:02:28:09 – 00:02:30:03
can cause harm.
00:02:30:03 – 00:02:31:16
So with VR, I
00:02:38:23 – 00:02:41:20
so with VR, AI, for example,
00:02:41:20 – 00:02:43:04
we have video,
00:02:43:04 – 00:02:44:06
remote
00:02:44:06 – 00:02:45:08
video, remote interpreters
00:02:45:08 – 00:02:47:07
that come up on the screen.
00:02:47:07 – 00:02:50:07
And sometimes it can be a great idea
00:02:50:17 – 00:02:52:04
because, you know,
00:02:52:04 – 00:02:55:04
when it first came out, we saw that
00:02:55:16 – 00:02:58:06
there was a lot of freezing.
00:02:58:06 – 00:03:00:23
Sometimes there might be challenges
00:03:00:23 – 00:03:01:20
internally, like,
00:03:01:20 – 00:03:02:19
let’s say
00:03:02:19 – 00:03:03:14
things are going on
00:03:03:14 – 00:03:06:02
in that room that cause issues.
00:03:06:02 – 00:03:06:20
Various issues
00:03:06:20 – 00:03:09:20
would happen with this technology.
00:03:09:21 – 00:03:13:01
Now, if you think about R2,
00:03:13:01 – 00:03:15:00
we think about automatic interpreting
00:03:15:00 – 00:03:16:07
or artificial intelligence
00:03:16:07 – 00:03:17:23
and interpreting.
00:03:17:23 – 00:03:19:11
Are we ready for that?
00:03:19:11 – 00:03:20:20
How will that impact
00:03:20:20 – 00:03:22:19
the greater community?
00:03:22:19 – 00:03:24:01
What will be the community’s
00:03:24:01 – 00:03:26:03
view of this?
00:03:26:03 – 00:03:28:07
So last fall
00:03:28:07 – 00:03:31:07
we hosted three webinars
00:03:31:10 – 00:03:33:23
and those webinars happened here
00:03:33:23 – 00:03:35:10
at Brown University.
00:03:35:10 – 00:03:38:03
They were through Zoom
00:03:38:03 – 00:03:40:13
and we had a panel
00:03:40:13 – 00:03:41:21
and we had discussions
00:03:41:21 – 00:03:43:00
and we gathered the view
00:03:43:00 – 00:03:44:12
of multiple people.
00:03:44:12 – 00:03:46:07
We also had deaf community members
00:03:46:07 – 00:03:48:02
who came in to watch
00:03:48:02 – 00:03:49:18
and make comments
00:03:49:18 – 00:03:51:23
and talk about their perspective.
00:03:53:17 – 00:03:56:12
So in that discussion we saw that
00:03:56:12 – 00:03:58:16
there was a lot of rich information
00:03:58:16 – 00:04:00:11
that was shared
00:04:00:11 – 00:04:03:11
and the team, the advisory group,
00:04:04:03 – 00:04:07:04
decided to work to analyze
00:04:07:14 – 00:04:08:24
that information that came
00:04:08:24 – 00:04:10:15
from the discussions.
00:04:10:15 – 00:04:11:17
And right away,
00:04:11:17 – 00:04:14:08
we knew we had a lot of rich content
00:04:14:08 – 00:04:15:15
and that we would be able
00:04:15:15 – 00:04:17:05
to take that content
00:04:17:05 – 00:04:19:01
and add it to the report
00:04:19:01 – 00:04:20:03
so that we could get
00:04:20:03 – 00:04:22:23
a general understanding of what fire
00:04:22:23 – 00:04:25:00
I would look like.
00:04:25:00 – 00:04:28:06
And we had several different questions
00:04:28:06 – 00:04:30:00
that we put into a survey
00:04:30:00 – 00:04:32:18
and we sent those out and we were able
00:04:32:18 – 00:04:36:01
to get those responses from the deaf.
00:04:36:08 – 00:04:36:23
Unfortunately,
00:04:36:23 – 00:04:39:08
we did not have a survey in ASL,
00:04:39:08 – 00:04:41:06
but we knew that we needed
00:04:41:06 – 00:04:42:14
to have these discussions
00:04:42:14 – 00:04:44:03
in order to gather the information
00:04:44:03 – 00:04:45:14
we were looking for.
00:04:45:14 – 00:04:46:03
Let’s go ahead and
00:04:46:03 – 00:04:47:00
head to the next slide.
00:04:53:03 – 00:04:55:06
So
00:04:55:06 – 00:04:55:17
I’d like
00:04:55:17 – 00:04:56:12
to introduce
00:04:56:12 – 00:04:59:12
you to other members of our team.
00:04:59:15 – 00:05:02:07
We all participated together
00:05:02:07 – 00:05:03:15
in gathering this research
00:05:03:15 – 00:05:06:05
and we’ve been working hard as a group
00:05:06:05 – 00:05:07:10
to get that information.
00:05:07:10 – 00:05:08:24
And these lovely people here
00:05:08:24 – 00:05:10:09
volunteer their time.
00:05:10:09 – 00:05:12:01
I’d like to introduce you to them now
00:05:12:01 – 00:05:14:03
so that you can get to know them a bit.
00:05:14:03 – 00:05:15:13
And they will also be talking
00:05:15:13 – 00:05:17:09
about the report today.
00:05:17:09 – 00:05:19:18
So
00:05:19:18 – 00:05:22:16
let me go ahead and pass it over to you.
00:05:22:16 – 00:05:25:09
Let’s start with Theresa.
00:05:25:09 – 00:05:25:17
Tracey,
00:05:25:17 – 00:05:28:07
if you’d like to introduce yourself.
00:05:28:07 – 00:05:28:17
Sure.
00:05:28:17 – 00:05:30:13
Good morning, everyone.
00:05:30:13 – 00:05:33:08
My name is Theresa Blake.
00:05:33:08 – 00:05:36:08
Maya Burke
00:05:36:09 – 00:05:38:09
and I work at Gallaudet University.
00:05:38:09 – 00:05:40:12
I’m a professor of philosophy
00:05:40:12 – 00:05:41:24
and my research
00:05:41:24 – 00:05:44:11
is specifically in ethics.
00:05:44:11 – 00:05:47:11
And it’s the ethical application
00:05:47:15 – 00:05:48:21
to technology.
00:05:48:21 – 00:05:50:09
And I’m thrilled to be here with you
00:05:50:09 – 00:05:50:22
all today.
00:05:50:22 – 00:05:52:00
I’m looking forward
00:05:52:00 – 00:05:53:17
to discussing the webinar
00:05:53:17 – 00:05:54:24
and having other discussions
00:05:54:24 – 00:05:56:03
with all of you today.
00:05:56:03 – 00:05:57:03
And I did like
00:05:57:03 – 00:05:58:05
I would like to add
00:05:58:05 – 00:06:01:05
that in terms of visual description,
00:06:01:11 – 00:06:02:10
I am a middle aged
00:06:02:10 – 00:06:06:03
woman with an olive, with olive skin
00:06:06:12 – 00:06:08:24
and also I have brown eyes,
00:06:08:24 – 00:06:09:11
I’m wearing
00:06:09:11 – 00:06:11:06
glasses and I have a brown
00:06:11:06 – 00:06:12:00
I have brown hair
00:06:12:00 – 00:06:13:13
and my hair is in a bun today
00:06:13:13 – 00:06:16:09
and I’m wearing a gray sweater
00:06:18:04 – 00:06:20:20
and I’m here in my office at Gallaudet.
00:06:20:20 – 00:06:22:22
Thank you, Theresa.
00:06:22:22 – 00:06:24:20
This next, let’s have Jeff.
00:06:24:20 – 00:06:25:02
Jeff,
00:06:25:02 – 00:06:26:15
would you like to introduce yourself?
00:06:26:15 – 00:06:29:04
Hello, My name is Jeff Schall
00:06:29:04 – 00:06:32:04
and I am working.
00:06:32:06 – 00:06:35:06
I work to develop
00:06:35:06 – 00:06:38:02
AI for the deaf and hard of hearing.
00:06:38:02 – 00:06:41:08
And in terms of a visual description,
00:06:41:17 – 00:06:43:00
I am wearing a white
00:06:43:00 – 00:06:44:21
and black plaid shirt.
00:06:44:21 – 00:06:47:15
I have facial hair and brown eyes
00:06:47:15 – 00:06:49:06
and I’m here.
00:06:49:06 – 00:06:50:10
My office is a background
00:06:50:10 – 00:06:51:17
and I work for go sign.
00:06:51:17 – 00:06:52:24
I
00:06:52:24 – 00:06:55:24
next will have Holly.
00:06:56:03 – 00:06:56:22
Yes, hello.
00:06:56:22 – 00:06:58:11
Good morning.
00:06:58:11 – 00:07:00:15
My name is Holly.
00:07:00:15 – 00:07:03:07
Last name is Jackson
00:07:03:07 – 00:07:06:07
and visual description.
00:07:06:16 – 00:07:09:16
I am an African-American black female.
00:07:10:04 – 00:07:14:02
I have light skin and I have curly
00:07:14:02 – 00:07:17:02
natural hair today.
00:07:17:19 – 00:07:18:12
And
00:07:19:11 – 00:07:22:11
I have a dark Navy
00:07:22:11 – 00:07:25:11
blue suit jacket on.
00:07:26:10 – 00:07:29:09
And I the shirt, my Navy blue
00:07:29:09 – 00:07:29:22
suit
00:07:29:22 – 00:07:32:22
jacket has light white stripes on it.
00:07:33:08 – 00:07:36:02
And I have a shirt beneath my jacket,
00:07:36:02 – 00:07:39:07
and it is a light blue tan
00:07:40:19 – 00:07:42:17
lace
00:07:42:17 – 00:07:43:20
top.
00:07:43:20 – 00:07:45:11
That’s the design.
00:07:45:11 – 00:07:48:14
And my background today
00:07:49:18 – 00:07:52:15
is light gray
00:07:52:15 – 00:07:54:17
plane, light gray background.
00:07:54:17 – 00:07:57:02
And I’m an interpreter.
00:07:57:02 – 00:07:59:09
I’m a hearing interpreter
00:07:59:09 – 00:08:01:15
and also an educator
00:08:01:15 – 00:08:04:08
and educator of ASL and interpreting.
00:08:04:08 – 00:08:06:11
I work for any ASL program
00:08:06:11 – 00:08:09:23
interpreting program, and also I am here
00:08:11:03 – 00:08:13:19
for Naomi,
00:08:13:19 – 00:08:15:13
the representation of Niobe,
00:08:15:13 – 00:08:18:15
the Atlanta chapter, and
00:08:18:15 – 00:08:21:15
I serve as the secretary for the board.
00:08:21:22 – 00:08:23:15
That’s my position this year.
00:08:23:15 – 00:08:24:13
Thank you very much.
00:08:24:13 – 00:08:26:13
I’m happy to be here.
00:08:26:13 – 00:08:27:22
Thank you, Holly.
00:08:27:22 – 00:08:30:15
And last but not least, Anne Marie.
00:08:32:10 – 00:08:35:10
Hello, I’m Anne Marie.
00:08:35:19 – 00:08:37:15
Last name is Killian,
00:08:37:15 – 00:08:40:15
and this is my signed name,
00:08:40:18 – 00:08:43:19
and I am the CEO for
00:08:45:23 – 00:08:49:16
Tie Access and visual description
00:08:49:16 – 00:08:50:14
is that I’m a white
00:08:50:14 – 00:08:53:19
female, middle aged with medium
00:08:53:19 – 00:08:54:24
length hair,
00:08:54:24 – 00:08:56:19
brown hair, and I’m wearing glasses.
00:08:56:19 – 00:08:57:17
Today
00:08:57:17 – 00:08:59:01
I have on a suit jacket
00:08:59:01 – 00:09:00:18
that is black with a purple shirt
00:09:00:18 – 00:09:01:23
beneath it.
00:09:01:23 – 00:09:03:11
And in the background
00:09:03:11 – 00:09:06:11
you can see my dining room table
00:09:06:13 – 00:09:09:21
and blacks, black chairs and
00:09:12:08 – 00:09:13:13
you might see two dogs
00:09:13:13 – 00:09:14:19
running around in the background.
00:09:14:19 – 00:09:16:23
If that happens, I apologize in advance.
00:09:16:23 – 00:09:18:09
Like everyone else.
00:09:18:09 – 00:09:19:13
I’m thrilled to be here today.
00:09:19:13 – 00:09:21:02
Thank you.
00:09:21:02 – 00:09:23:12
Great.
00:09:23:12 – 00:09:26:12
So let’s go to the next slide.
00:09:28:17 – 00:09:29:11
So today
00:09:29:11 – 00:09:31:07
we will be talking about
00:09:31:07 – 00:09:32:14
multiple things.
00:09:32:14 – 00:09:34:04
And during these presentations,
00:09:34:04 – 00:09:36:05
we’ll go in depth about our studies
00:09:36:05 – 00:09:38:04
and what we have found through analyzing
00:09:38:04 – 00:09:39:02
this data
00:09:39:02 – 00:09:40:13
will be sharing with you
00:09:40:13 – 00:09:42:17
this information today.
00:09:42:17 – 00:09:44:04
The first thing we’re going to be doing
00:09:44:04 – 00:09:45:08
is identifying
00:09:45:08 – 00:09:47:21
three critical impact areas.
00:09:49:00 – 00:09:51:02
And so these impact
00:09:51:02 – 00:09:52:16
areas are quite important.
00:09:52:16 – 00:09:54:05
We’ll be talking more in depth
00:09:54:05 – 00:09:56:12
and describing what they are for you.
00:09:56:12 – 00:09:59:12
Secondly,
00:09:59:12 – 00:10:00:17
with this analysis
00:10:00:17 – 00:10:03:17
and with this research in our webinars,
00:10:04:07 – 00:10:06:14
we were able to go through the data
00:10:06:14 – 00:10:08:00
and that data helped us
00:10:08:00 – 00:10:11:06
to build a better understanding of what
00:10:11:06 – 00:10:12:23
the deaf communities perspective
00:10:12:23 – 00:10:14:18
and experience has been
00:10:14:18 – 00:10:17:17
and their experiences with interpreters
00:10:17:17 – 00:10:19:04
and the harms that have happened
00:10:19:04 – 00:10:20:18
and the way that these experiences
00:10:20:18 – 00:10:21:08
have impacted
00:10:21:08 – 00:10:22:07
their life
00:10:22:07 – 00:10:23:13
in terms of access,
00:10:23:13 – 00:10:25:11
in terms of communication access.
00:10:25:11 – 00:10:26:20
And so it’s very important
00:10:26:20 – 00:10:28:17
to have this deaf community perspective
00:10:28:17 – 00:10:29:20
so that we can understand
00:10:29:20 – 00:10:30:19
what they’ve been through
00:10:30:19 – 00:10:32:00
when it comes to
00:10:32:00 – 00:10:34:01
their experiences in interpreting
00:10:34:01 – 00:10:36:05
and where harm has happened.
00:10:36:05 – 00:10:36:19
Often
00:10:36:19 – 00:10:38:17
that is an experience
00:10:38:17 – 00:10:40:16
that is common in our communities.
00:10:40:16 – 00:10:42:23
So we would like to mitigate those harms
00:10:42:23 – 00:10:47:02
and ensure that if we do put out
00:10:47:02 – 00:10:48:00
new technology,
00:10:48:00 – 00:10:49:08
that it’s going to be we’re going
00:10:49:08 – 00:10:50:16
to be mindful of those harms
00:10:50:16 – 00:10:52:06
that have been experience.
00:10:52:06 – 00:10:53:14
Third, we’re going to talk
00:10:53:14 – 00:10:55:10
about the value of the big picture
00:10:55:10 – 00:10:57:08
lens on possibilities.
00:10:57:08 – 00:10:58:12
We’ll talk about
00:10:58:12 – 00:11:00:01
what it looks like to do right
00:11:00:01 – 00:11:02:11
and prevent possible disaster.
00:11:02:11 – 00:11:05:23
We want to make sure that we are looking
00:11:05:23 – 00:11:08:12
through a lens where we are
00:11:08:12 – 00:11:10:02
showing care and concern
00:11:10:02 – 00:11:11:24
about the future of the community
00:11:11:24 – 00:11:13:10
and taking all of these things
00:11:13:10 – 00:11:15:00
into account. Next slide.
00:11:24:19 – 00:11:26:11
So for today
00:11:26:11 – 00:11:29:03
in this webinar, like you mentioned,
00:11:29:03 – 00:11:30:06
we had a panel
00:11:30:06 – 00:11:31:17
and we had
00:11:31:17 – 00:11:32:11
today will be going
00:11:32:11 – 00:11:34:06
through different presentations
00:11:34:06 – 00:11:37:10
and we’ll be going through the report
00:11:37:10 – 00:11:39:22
and having discussions about that.
00:11:39:22 – 00:11:42:20
So we will be following this order,
00:11:42:20 – 00:11:45:20
as you see outlined here and the slide
00:11:46:24 – 00:11:48:05
we will be discussing
00:11:48:05 – 00:11:51:05
ethics and fairness.
00:11:52:02 – 00:11:54:17
We will be discussing how this research
00:11:54:17 – 00:11:55:24
was studied
00:11:55:24 – 00:11:58:24
and our approach to the research.
00:11:59:06 – 00:12:02:02
We will be talking about what we found.
00:12:02:02 – 00:12:04:10
So our findings,
00:12:04:10 – 00:12:07:13
we will also be discussing the results
00:12:08:03 – 00:12:11:07
and outcomes of our research.
00:12:12:02 – 00:12:13:17
And also we will be talking
00:12:13:17 – 00:12:14:18
about concerns
00:12:14:18 – 00:12:17:18
as it relates to the deaf community.
00:12:18:07 – 00:12:20:14
We will be also discussing techno
00:12:20:14 – 00:12:23:14
technology and the quality
00:12:25:05 – 00:12:27:23
and we will be asking,
00:12:27:23 – 00:12:29:10
are we ready for this?
00:12:29:10 – 00:12:30:20
Are we ready for eye,
00:12:32:06 – 00:12:33:00
for eye
00:12:33:00 – 00:12:36:00
and sign language to come together?
00:12:36:20 – 00:12:37:18
Are we ready?
00:12:37:18 – 00:12:40:18
And if we are, what does that look like?
00:12:41:18 – 00:12:44:18
What kind of risks are involved?
00:12:46:00 – 00:12:48:11
We need to be proactive
00:12:48:11 – 00:12:50:19
in understanding
00:12:50:19 – 00:12:52:09
and predicting those risks
00:12:52:09 – 00:12:53:21
so that we can mitigate
00:12:53:21 – 00:12:56:21
or resolve them before they occur.
00:12:59:18 – 00:13:02:08
We will also discuss the future
00:13:02:08 – 00:13:04:17
and what we can anticipate
00:13:04:17 – 00:13:08:02
and what we can recommend for
00:13:09:06 – 00:13:11:03
any anyone
00:13:11:03 – 00:13:13:22
who is going to be working in relation
00:13:13:22 – 00:13:15:08
to this topic.
00:13:15:08 – 00:13:18:08
All of us here today are impacted
00:13:18:08 – 00:13:22:23
or affected by this topic and many people
00:13:22:23 – 00:13:23:15
who are not here
00:13:23:15 – 00:13:24:15
today,
00:13:24:15 – 00:13:27:15
the entire community that we represent
00:13:27:22 – 00:13:30:22
and also work with.
00:13:32:05 – 00:13:34:17
So now
00:13:34:17 – 00:13:36:11
Teresa
00:13:36:11 – 00:13:39:11
will go ahead and begin this discussion
00:13:39:13 – 00:13:40:20
and we’re going to start off again
00:13:40:20 – 00:13:42:09
with ethics and fairness.
00:13:42:09 – 00:13:43:18
Teresa, I’ll turn it over to you.
00:13:49:06 – 00:13:50:09
one moment.
00:13:50:09 – 00:13:53:09
Let me make sure I have this up.
00:13:54:11 – 00:13:55:09
I always have to make sure
00:13:55:09 – 00:13:58:05
I have my spotlight on so that I’m seen.
00:13:58:05 – 00:13:58:17
Okay.
00:13:58:17 – 00:14:02:12
So ethics, ethics and fairness,
00:14:03:01 – 00:14:04:23
what emphasis
00:14:04:23 – 00:14:06:14
and what are we looking at here
00:14:06:14 – 00:14:07:18
with ethics and fairness?
00:14:07:18 – 00:14:09:12
What do we want to avoid?
00:14:09:12 – 00:14:11:00
We don’t want to cause harm.
00:14:11:00 – 00:14:13:17
How do we reduce the cause of harm?
00:14:13:17 – 00:14:15:02
And what does harm mean
00:14:15:02 – 00:14:16:14
to the deaf community
00:14:16:14 – 00:14:18:16
and deaf individuals?
00:14:18:16 – 00:14:21:22
Harm in general, to humanity?
00:14:21:22 – 00:14:23:12
We want to avoid.
00:14:23:12 – 00:14:27:00
So we have two
00:14:27:00 – 00:14:30:00
topics really
00:14:30:00 – 00:14:32:08
coexisting here ethics and fairness.
00:14:32:08 – 00:14:35:13
So what you’ll notice here,
00:14:35:20 – 00:14:38:24
we are concerned with controlling bias
00:14:39:24 – 00:14:42:00
and we want to
00:14:42:00 – 00:14:43:10
assign responsibility
00:14:43:10 – 00:14:46:10
or accountability, rather, as it
00:14:46:22 – 00:14:49:11
relates to AI to avoid harm.
00:14:51:13 – 00:14:52:04
The second
00:14:52:04 – 00:14:52:17
point I like
00:14:52:17 – 00:14:53:00
to make
00:14:53:00 – 00:14:55:05
is that we need to be very clear
00:14:55:05 – 00:14:58:14
and transparent with our documentation
00:14:58:14 – 00:15:00:02
about who is accountable
00:15:00:02 – 00:15:02:00
in the design of AI
00:15:02:00 – 00:15:03:24
and the development of AI
00:15:03:24 – 00:15:06:22
and the application and evaluation.
00:15:06:22 – 00:15:09:07
Everything related to A.I.
00:15:09:07 – 00:15:11:19
who is responsible for this portion
00:15:11:19 – 00:15:13:11
ethically?
00:15:13:11 – 00:15:16:11
Next slide.
00:15:20:01 – 00:15:20:13
Okay.
00:15:20:13 – 00:15:22:05
So we spoke about ethics,
00:15:22:05 – 00:15:23:01
and now I want to talk
00:15:23:01 – 00:15:25:08
a little bit about fairness.
00:15:25:08 – 00:15:28:19
How we in our society
00:15:28:20 – 00:15:31:20
have developed the meaning behind this.
00:15:31:20 – 00:15:35:12
We have a lot of negative biases,
00:15:35:16 – 00:15:36:08
and we also have
00:15:36:08 – 00:15:38:18
positive biases in society.
00:15:38:18 – 00:15:39:18
But what we want to do
00:15:39:18 – 00:15:42:02
is make sure that in air
00:15:42:02 – 00:15:44:07
we want to reflect the best of society
00:15:44:07 – 00:15:45:24
and not the worst.
00:15:45:24 – 00:15:46:20
So with that being said,
00:15:46:20 – 00:15:48:05
we want to eliminate bias
00:15:48:05 – 00:15:49:10
and we want to eliminate
00:15:49:10 – 00:15:50:24
eliminate favoritism
00:15:50:24 – 00:15:52:19
with measurable results.
00:15:52:19 – 00:15:56:12
And also we want to be able to observe
00:15:56:12 – 00:15:59:21
how people are interacting with that.
00:15:59:21 – 00:16:01:13
We have statistics and evidence
00:16:01:13 – 00:16:03:01
related to the fairness,
00:16:03:01 – 00:16:04:10
and we need to use that
00:16:04:10 – 00:16:06:10
fairness evidence based.
00:16:07:15 – 00:16:10:15
Next slide.
00:16:14:17 – 00:16:15:13
Okay.
00:16:15:13 – 00:16:17:01
This one is a little bit interesting
00:16:17:01 – 00:16:18:09
because it’s a little bit
00:16:18:09 – 00:16:21:09
of a mixture of English and
00:16:21:09 – 00:16:23:09
written English and sign language.
00:16:23:09 – 00:16:26:09
So what we see here is the word A.I.
00:16:26:09 – 00:16:29:09
by A.I.. A.I.
00:16:29:18 – 00:16:30:20
Times, A.I.
00:16:30:20 – 00:16:32:21
on the screen in a mathematical look.
00:16:32:21 – 00:16:34:09
But we have a few different ways
00:16:34:09 – 00:16:37:10
that we are expressing this concept Now.
00:16:37:10 – 00:16:38:13
If we’re signing it,
00:16:38:13 – 00:16:40:13
you may see a sign A.I.
00:16:40:13 – 00:16:42:05
squared. Okay?
00:16:42:05 – 00:16:43:01
But when we say
00:16:43:01 – 00:16:46:01
that we’re referring to automated
00:16:46:01 – 00:16:46:19
or automatic
00:16:46:19 – 00:16:48:00
interpreting by
00:16:48:00 – 00:16:50:02
artificial intelligence, A.I.
00:16:50:02 – 00:16:51:05
by A.I..
00:16:51:05 – 00:16:51:23
Now, sometimes
00:16:51:23 – 00:16:55:13
you may see people sign a AI by A.I.,
00:16:55:24 – 00:16:57:05
and it’s the same concept.
00:16:57:05 – 00:16:58:21
It means the same thing.
00:16:58:21 – 00:17:00:00
Now, in written form.
00:17:00:00 – 00:17:03:00
We’ll see A.I., X, A.I..
00:17:03:01 – 00:17:04:11
Those are the three ways you’ll see it.
00:17:04:11 – 00:17:06:08
But the main point is understanding
00:17:06:08 – 00:17:07:03
that we’re referring
00:17:07:03 – 00:17:10:12
to automatic interpreting by artificial
00:17:11:09 – 00:17:12:04
intelligence.
00:17:15:02 – 00:17:16:07
Now, with that being said,
00:17:16:07 – 00:17:19:07
I’m going to turn this over to
00:17:20:13 – 00:17:21:18
I’m sorry, I don’t remember
00:17:21:18 – 00:17:22:23
who’s next on here.
00:17:22:23 – 00:17:25:23
Let me work
00:17:26:13 – 00:17:29:13
just.
00:17:31:12 – 00:17:31:23
Hello.
00:17:31:23 – 00:17:34:23
Okay, so
00:17:35:24 – 00:17:37:12
as Teresa mentioned,
00:17:37:12 – 00:17:39:23
we hosted three webinars.
00:17:39:23 – 00:17:41:12
And during those webinars,
00:17:41:12 – 00:17:42:06
we invited
00:17:42:06 – 00:17:43:09
so many of the deaf
00:17:43:09 – 00:17:46:09
community members to participate with us.
00:17:46:19 – 00:17:48:05
Those webinars,
00:17:48:05 – 00:17:50:22
we discussed a variety of topics
00:17:50:22 – 00:17:51:23
and issues
00:17:51:23 – 00:17:54:11
covering AI and interpreting
00:17:54:11 – 00:17:55:18
and how they relate.
00:17:55:18 – 00:17:57:14
As Teresa mentioned, A.I.
00:17:57:14 – 00:17:59:21
Squared is how we were referring to A.I.
00:17:59:21 – 00:18:00:21
by A.I..
00:18:00:21 – 00:18:02:22
There was so much
00:18:02:22 – 00:18:05:09
great dialog conversation,
00:18:05:09 – 00:18:07:07
ideas and issues
00:18:07:07 – 00:18:09:18
discussed during these webinars.
00:18:09:18 – 00:18:12:16
Now we recorded these webinars
00:18:12:16 – 00:18:15:16
and we made a transcript of them
00:18:15:22 – 00:18:17:18
and we went through these transcripts
00:18:17:18 – 00:18:19:01
with a fine tooth comb
00:18:19:01 – 00:18:20:07
and we looked at them
00:18:20:07 – 00:18:21:12
and looked at patterns
00:18:21:12 – 00:18:23:04
that arose in each
00:18:23:04 – 00:18:24:04
discussion, different
00:18:24:04 – 00:18:25:11
themes that popped up
00:18:25:11 – 00:18:26:21
throughout the entirety
00:18:26:21 – 00:18:28:02
of these webinars.
00:18:28:02 – 00:18:29:11
And we wanted to make sure
00:18:29:11 – 00:18:31:12
that we were able to pinpoint
00:18:31:12 – 00:18:32:21
and really understand
00:18:32:21 – 00:18:34:21
what issues are most prevalent
00:18:34:21 – 00:18:36:10
to the community at large.
00:18:38:02 – 00:18:39:17
We did
00:18:39:17 – 00:18:42:08
a time analysis as well
00:18:42:08 – 00:18:44:22
and we looked at how frequently
00:18:44:22 – 00:18:46:14
different themes
00:18:46:14 – 00:18:48:18
popped up in said webinars
00:18:48:18 – 00:18:50:13
and we found
00:18:50:13 – 00:18:52:07
our findings
00:18:52:07 – 00:18:53:06
are shown in the next slide.
00:18:53:06 – 00:18:54:01
And if you’re interested
00:18:54:01 – 00:18:55:09
in more detailed information,
00:18:55:09 – 00:18:56:19
please read our report.
00:18:56:19 – 00:18:59:22
It goes into a varied in depth
00:19:01:05 – 00:19:03:02
reporting of our findings
00:19:03:02 – 00:19:05:00
and that is available online as well.
00:19:05:00 – 00:19:08:00
Next slide, please.
00:19:10:20 – 00:19:14:12
So this is a snapshot of our findings.
00:19:14:18 – 00:19:16:01
We put down
00:19:16:01 – 00:19:16:19
all the themes
00:19:16:19 – 00:19:17:12
that popped up
00:19:17:12 – 00:19:20:08
most frequently in our webinars,
00:19:20:08 – 00:19:21:22
and we categorized those
00:19:21:22 – 00:19:25:22
into three different areas of study.
00:19:27:00 – 00:19:30:08
The first area is related
00:19:30:08 – 00:19:33:08
to results and outcomes,
00:19:34:01 – 00:19:35:03
and we’ll talk about this
00:19:35:03 – 00:19:36:09
more in the next slide.
00:19:36:09 – 00:19:39:09
We’re focused more on discussing about
00:19:39:20 – 00:19:42:19
what kind of society,
00:19:42:19 – 00:19:44:24
what kind of societal results
00:19:44:24 – 00:19:48:20
will arise from the impact of AI by AI.
00:19:50:08 – 00:19:53:04
Our next theme was readiness.
00:19:53:04 – 00:19:56:04
How ready is the community at large?
00:19:56:21 – 00:20:00:05
How ready are stakeholders for
00:20:01:22 – 00:20:03:14
this technology?
00:20:03:14 – 00:20:04:19
We looked at the feedback
00:20:04:19 – 00:20:06:03
and the requirements and things
00:20:06:03 – 00:20:07:04
that need to happen
00:20:07:04 – 00:20:08:15
for the community at large
00:20:08:15 – 00:20:10:19
in this technological realm.
00:20:10:19 – 00:20:13:05
But as you can see on here,
00:20:13:05 – 00:20:16:05
by over half and half,
00:20:16:08 – 00:20:17:09
the biggest issue
00:20:17:09 – 00:20:19:01
that came up most frequently
00:20:19:01 – 00:20:20:08
and most prevalent on people’s
00:20:20:08 – 00:20:22:17
minds was technological quality.
00:20:22:17 – 00:20:24:08
What type of regulations
00:20:24:08 – 00:20:27:19
and standards are required to minimize
00:20:28:03 – 00:20:31:03
the reduce or the possibility of harm?
00:20:31:15 – 00:20:33:08
And how do we maximize
00:20:33:08 – 00:20:35:01
the potential benefits
00:20:35:01 – 00:20:36:17
of this technology?
00:20:36:17 – 00:20:37:22
Now, for this,
00:20:37:22 – 00:20:39:14
we went into more in-depth
00:20:39:14 – 00:20:42:03
research in our report,
00:20:42:03 – 00:20:43:24
and I’m going to turn it
00:20:43:24 – 00:20:44:20
over to Ann Marie.
00:20:44:20 – 00:20:45:12
If she can.
00:20:45:12 – 00:20:46:10
She can discuss this
00:20:46:10 – 00:20:47:08
a little bit more in depth.
00:20:49:05 – 00:20:50:08
Thank you so much, Jeff.
00:20:50:08 – 00:20:53:08
I really appreciate it.
00:20:53:08 – 00:20:54:06
The first thing that we’re going
00:20:54:06 – 00:20:55:07
to look at, again,
00:20:55:07 – 00:20:59:03
as Jeff mentioned in his synopsis
00:20:59:03 – 00:21:01:13
of the percentages that we looked at,
00:21:01:13 – 00:21:03:06
it is staggering.
00:21:03:06 – 00:21:06:06
The biggest section was on technological
00:21:06:20 – 00:21:09:20
aspects, but
00:21:09:20 – 00:21:13:02
I’m looking at the desired results
00:21:13:02 – 00:21:14:15
and outcomes as well,
00:21:14:15 – 00:21:16:07
and it was a very hot topic
00:21:16:07 – 00:21:18:18
in the discussions in the webinar.
00:21:18:18 – 00:21:19:20
It was really important
00:21:19:20 – 00:21:22:20
to emphasize that as well.
00:21:24:05 – 00:21:25:08
Now, more than one
00:21:25:08 – 00:21:26:17
third of the discussions
00:21:26:17 – 00:21:29:17
regarding this were really focused on
00:21:30:02 – 00:21:33:02
these critical areas of control,
00:21:34:16 – 00:21:36:19
control
00:21:36:19 – 00:21:39:16
by the language community
00:21:39:16 – 00:21:40:18
and the authority
00:21:40:18 – 00:21:42:24
who is involved in the decision making,
00:21:42:24 – 00:21:44:15
who is involved in the device
00:21:44:15 – 00:21:45:23
development of design
00:21:45:23 – 00:21:48:02
and the impact the A.I. is going to have.
00:21:49:16 – 00:21:50:24
The biggest concern
00:21:50:24 – 00:21:53:24
was really regarding the legality,
00:21:54:01 – 00:21:58:20
the structure and the synthesis of this.
00:21:59:03 – 00:22:00:20
And it was critical.
00:22:00:20 – 00:22:02:00
It was impressive
00:22:02:00 – 00:22:05:00
to see how much the deaf leaders
00:22:05:01 – 00:22:07:01
felt that,
00:22:07:01 – 00:22:08:03
you know, in history
00:22:08:03 – 00:22:09:20
in their based on their experience,
00:22:09:20 – 00:22:11:18
the deaf have not always
00:22:11:18 – 00:22:13:17
had a voice at the table,
00:22:13:17 – 00:22:15:05
whether it was in the development
00:22:15:05 – 00:22:17:15
of different views or different
00:22:17:15 – 00:22:19:01
processes.
00:22:19:01 – 00:22:20:01
The deaf perspective
00:22:20:01 – 00:22:20:15
has often
00:22:20:15 – 00:22:21:07
been missing
00:22:21:07 – 00:22:22:18
and it’s really critical
00:22:22:18 – 00:22:23:18
that they are included
00:22:23:18 – 00:22:24:17
in the conversation.
00:22:24:17 – 00:22:26:03
And what does that look like?
00:22:26:03 – 00:22:27:19
What does the deaf representation
00:22:27:19 – 00:22:28:06
look like
00:22:28:06 – 00:22:29:15
in the authoritative process
00:22:29:15 – 00:22:32:15
of developing these standards?
00:22:32:21 – 00:22:34:17
Also,
00:22:34:17 – 00:22:38:07
it is very important and very stringent
00:22:38:07 – 00:22:42:08
that any violation of these processes,
00:22:44:14 – 00:22:45:09
you know, if
00:22:45:09 – 00:22:47:01
these processes are violated,
00:22:47:01 – 00:22:48:20
what type of ramifications
00:22:48:20 – 00:22:49:15
does that have?
00:22:49:15 – 00:22:51:11
That’s been a very big discussion
00:22:51:11 – 00:22:53:07
topic as well. Next slide.
00:23:03:20 – 00:23:05:02
Now you see this next slide.
00:23:05:02 – 00:23:06:06
We’ve also included
00:23:06:06 – 00:23:09:09
some concerns based on our webinars
00:23:10:01 – 00:23:11:21
about control.
00:23:11:21 – 00:23:13:16
There are two levels of control,
00:23:13:16 – 00:23:17:06
individual control and cultural groups,
00:23:17:12 – 00:23:18:17
their control
00:23:18:17 – 00:23:19:20
and the impact
00:23:19:20 – 00:23:21:04
and the protections
00:23:21:04 – 00:23:23:16
needed for each of these groups.
00:23:23:16 – 00:23:25:02
We need to be sensitive to those,
00:23:25:02 – 00:23:28:02
especially with children.
00:23:29:21 – 00:23:31:19
The focus of the community at large.
00:23:31:19 – 00:23:33:16
The biggest priority here
00:23:33:16 – 00:23:37:08
was to make sure that the legal framework
00:23:39:01 – 00:23:41:00
is established with ABI.
00:23:41:00 – 00:23:44:00
I again, you know, it’s
00:23:44:12 – 00:23:48:11
one we’ve had to look at leaders involved
00:23:48:16 – 00:23:50:00
with the deaf community
00:23:50:00 – 00:23:51:17
involvement in research
00:23:51:17 – 00:23:54:17
and in development for these
00:23:55:02 – 00:23:55:20
technologies.
00:23:55:20 – 00:23:57:05
It’s very imperative
00:23:57:05 – 00:23:58:20
because it impacts their life
00:23:58:20 – 00:23:59:21
for the deaf community
00:23:59:21 – 00:24:00:10
at large,
00:24:00:10 – 00:24:00:23
for the deaf
00:24:00:23 – 00:24:01:20
blind community
00:24:01:20 – 00:24:02:18
as well,
00:24:02:18 – 00:24:04:14
for those that are losing their hearing
00:24:04:14 – 00:24:05:21
or hard of hearing
00:24:05:21 – 00:24:07:10
and learning sign,
00:24:07:10 – 00:24:09:14
this impacts all of them.
00:24:09:14 – 00:24:11:13
Again, it really can’t emphasize
00:24:11:13 – 00:24:12:17
this enough.
00:24:12:17 – 00:24:13:24
You know, the deaf
00:24:13:24 – 00:24:15:18
leadership needs to be involved
00:24:15:18 – 00:24:16:14
in the developmental
00:24:16:14 – 00:24:18:23
process of these platforms.
00:24:20:09 – 00:24:23:09
For I by
00:24:25:04 – 00:24:26:10
also it’s really interesting
00:24:26:10 – 00:24:27:04
to note
00:24:27:04 – 00:24:30:04
the importance of the concern that
00:24:30:24 – 00:24:33:24
I, I the data.
00:24:33:24 – 00:24:36:07
What is that data stored storage
00:24:36:07 – 00:24:37:01
look like?
00:24:37:01 – 00:24:39:10
Who is in conservatorship
00:24:39:10 – 00:24:40:20
of this storage?
00:24:40:20 – 00:24:41:21
What are the analytics
00:24:41:21 – 00:24:44:00
and the statistics looking at?
00:24:44:00 – 00:24:45:15
Where is that data held
00:24:45:15 – 00:24:47:00
and how is it protected?
00:24:47:00 – 00:24:49:10
That’s a very big discussion as well.
00:24:49:10 – 00:24:51:15
There’s a lot of concern about,
00:24:51:15 – 00:24:54:24
you know, hiring individuals
00:24:54:24 – 00:24:57:24
that could influence the development and
00:25:00:20 – 00:25:02:09
leadership of this.
00:25:02:09 – 00:25:03:13
That doesn’t always happen
00:25:03:13 – 00:25:05:00
to have deaf individuals
00:25:05:00 – 00:25:07:04
in this type of role
00:25:07:04 – 00:25:09:08
and involved in that process.
00:25:09:08 – 00:25:11:21
So will this process in the future
00:25:11:21 – 00:25:12:17
involve deaf
00:25:12:17 – 00:25:14:11
individuals in the beginning,
00:25:14:11 – 00:25:16:17
a foundational development of this,
00:25:16:17 – 00:25:18:16
or will they only bring them in for A
00:25:18:16 – 00:25:20:20
to perspective and that’s it.
00:25:20:20 – 00:25:22:00
So I think another good point
00:25:22:00 – 00:25:22:19
of discussion
00:25:22:19 – 00:25:23:08
as well,
00:25:23:08 – 00:25:24:17
who have the ownership and the
00:25:24:17 – 00:25:26:14
influence in this process.
00:25:29:22 – 00:25:31:20
You know, not only
00:25:31:20 – 00:25:33:19
hiring deaf consultants,
00:25:33:19 – 00:25:35:09
but all through the line,
00:25:35:09 – 00:25:36:01
through A.I.,
00:25:36:01 – 00:25:38:04
by having deaf
00:25:38:04 – 00:25:40:16
hands on the process is vital.
00:25:40:16 – 00:25:42:08
And I’m going to turn this over
00:25:42:08 – 00:25:45:08
to Jeff again.
00:25:49:08 – 00:25:51:20
Hi, it’s Jeff here again for you.
00:25:51:20 – 00:25:53:12
So
00:25:53:12 – 00:25:55:22
about the technological quality.
00:25:55:22 – 00:25:56:16
During the webinar,
00:25:56:16 – 00:25:58:19
we discussed this quite in depth
00:25:58:19 – 00:26:00:09
and how we can break down
00:26:00:09 – 00:26:03:10
this idea into several subcategories.
00:26:04:16 – 00:26:07:03
The first or the major subcategory
00:26:07:03 – 00:26:08:10
that a lot of people really
00:26:08:10 – 00:26:11:10
discussed in the webinar was data.
00:26:11:14 – 00:26:14:21
Many people agreed that the quality
00:26:15:05 – 00:26:16:19
and the diversity of the data
00:26:16:19 – 00:26:19:10
is key to the foundation of building A.I.
00:26:19:10 – 00:26:21:07
by A.I..
00:26:21:07 – 00:26:22:08
You know, with machine
00:26:22:08 – 00:26:25:08
learning in the community,
00:26:26:15 – 00:26:28:22
it’s garbage in, garbage out
00:26:28:22 – 00:26:30:06
more often than not.
00:26:30:06 – 00:26:32:12
And so what does that look like?
00:26:32:12 – 00:26:34:13
So if we have a model training
00:26:35:14 – 00:26:36:08
by that
00:26:36:08 – 00:26:38:03
with that data,
00:26:38:03 – 00:26:40:19
it will be the same in the output.
00:26:40:19 – 00:26:42:11
So
00:26:42:11 – 00:26:44:08
that leads us to our second
00:26:44:08 – 00:26:47:08
largest issue modeling.
00:26:47:23 – 00:26:50:09
So if we have the garbage in modeling,
00:26:50:09 – 00:26:51:07
it’s garbage out.
00:26:51:07 – 00:26:52:09
But people are looking at it
00:26:52:09 – 00:26:53:09
and saying, okay,
00:26:53:09 – 00:26:55:15
so what is the primary use?
00:26:55:15 – 00:26:57:11
What is the primary model?
00:26:57:11 – 00:26:59:07
What is what features are they using?
00:26:59:07 – 00:27:02:03
Are they focusing only on hand shape
00:27:02:03 – 00:27:02:18
or will
00:27:02:18 – 00:27:04:18
they also be including facial shape
00:27:04:18 – 00:27:05:08
and mouth
00:27:05:08 – 00:27:08:08
shapes, expressions and other key
00:27:08:15 – 00:27:11:15
contextual clues of the language?
00:27:12:09 – 00:27:14:22
How will we be able to evaluate
00:27:14:22 – 00:27:15:24
that model?
00:27:15:24 – 00:27:17:07
Can we decide
00:27:17:07 – 00:27:20:09
which model to use over another model,
00:27:20:21 – 00:27:23:21
and how do we use metrics
00:27:24:01 – 00:27:24:22
to decide
00:27:24:22 – 00:27:26:06
if that is the quality
00:27:26:06 – 00:27:27:20
that we need or not?
00:27:27:20 – 00:27:29:06
Many people really had
00:27:29:06 – 00:27:30:06
that on their minds
00:27:30:06 – 00:27:32:00
in the webinar discussions.
00:27:32:00 – 00:27:33:24
Now, throughout these discussions,
00:27:33:24 – 00:27:35:07
we also noticed the need
00:27:35:07 – 00:27:37:07
for deaf leadership
00:27:37:07 – 00:27:40:02
involvement and oversight as well.
00:27:40:02 – 00:27:44:01
It was key that that topic came up
00:27:44:01 – 00:27:44:24
many times.
00:27:44:24 – 00:27:46:13
We needed to set up
00:27:46:13 – 00:27:50:19
at least a minimum criteria for
00:27:52:00 – 00:27:54:15
the bidirectional interpreting for A.I.
00:27:54:15 – 00:27:55:13
by A.I..
00:27:55:13 – 00:27:58:13
Next slide, please.
00:28:01:16 – 00:28:04:18
Now many more subcategories
00:28:04:18 – 00:28:06:05
arose from this discussion,
00:28:06:05 – 00:28:08:02
and one of them was quality,
00:28:08:02 – 00:28:10:02
and it was very imperative
00:28:10:02 – 00:28:13:02
This topic popped up quite a bit
00:28:13:24 – 00:28:16:24
as well.
00:28:17:01 – 00:28:18:24
The participants all agreed
00:28:18:24 – 00:28:23:03
that during the process of gathering
00:28:23:13 – 00:28:26:18
data, storing data, using data,
00:28:27:05 – 00:28:29:24
sharing data, publishing
00:28:29:24 – 00:28:30:17
all of that,
00:28:30:17 – 00:28:31:18
and that process
00:28:31:18 – 00:28:33:19
needed to be quite transparent
00:28:33:19 – 00:28:34:23
so that we could understand
00:28:34:23 – 00:28:37:12
everything as it happened.
00:28:37:12 – 00:28:39:14
We need the opt in or opt out
00:28:39:14 – 00:28:41:07
option as well.
00:28:41:07 – 00:28:43:03
Also, discussing the ability
00:28:43:03 – 00:28:44:12
to withdrawal.
00:28:44:12 – 00:28:46:14
Our can fit in the future,
00:28:46:14 – 00:28:48:12
meaning if we opt into this
00:28:48:12 – 00:28:49:06
and then later
00:28:49:06 – 00:28:50:09
we change our minds and say,
00:28:50:09 – 00:28:50:18
you know what,
00:28:50:18 – 00:28:52:10
we don’t want to participate.
00:28:52:10 – 00:28:54:23
We have the option to withdraw
00:28:54:23 – 00:28:55:18
that consent.
00:28:57:10 – 00:28:59:04
Another topic that we discussed
00:28:59:04 – 00:28:59:22
quite in death
00:28:59:22 – 00:29:03:01
was the topic of safeguards.
00:29:03:23 – 00:29:05:12
Those are needed to be put in place
00:29:05:12 – 00:29:07:10
to minimize harm.
00:29:07:10 – 00:29:10:10
Many of the what ifs
00:29:10:13 – 00:29:12:18
that arose during our discussion
00:29:12:18 – 00:29:14:03
really related to this.
00:29:14:03 – 00:29:16:12
What if someone downloads this?
00:29:16:12 – 00:29:20:15
What if there is a breach of data?
00:29:20:21 – 00:29:23:14
What if the information is leaked?
00:29:23:14 – 00:29:25:19
What if the system crashes?
00:29:25:19 – 00:29:26:15
There were so many
00:29:26:15 – 00:29:28:09
what ifs and questions
00:29:28:09 – 00:29:30:09
that popped up in that discussion.
00:29:30:09 – 00:29:31:00
And again,
00:29:31:00 – 00:29:33:00
for more analysis of this, please
00:29:33:00 – 00:29:33:23
look at our report.
00:29:33:23 – 00:29:36:17
We have a lot of it detailed in there.
00:29:36:17 – 00:29:39:17
And we also talked about readiness.
00:29:40:10 – 00:29:42:08
And Theresa is going to talk about
00:29:42:08 – 00:29:44:19
that more in depth now.
00:29:44:19 – 00:29:47:19
I’ll hand it over to Theresa.
00:29:51:10 – 00:29:53:20
Okay.
00:29:53:20 – 00:29:56:20
So
00:29:57:08 – 00:29:59:20
in terms of readiness,
00:29:59:20 – 00:30:02:05
when we look at what that includes,
00:30:02:05 – 00:30:04:06
we have readiness
00:30:04:06 – 00:30:05:16
when it comes to technology,
00:30:05:16 – 00:30:09:03
but also there’s a discussion of society
00:30:09:10 – 00:30:11:12
and what our awareness
00:30:11:12 – 00:30:12:01
looks like
00:30:12:01 – 00:30:13:19
in terms of ethical issues
00:30:13:19 – 00:30:15:22
and ethical concerns.
00:30:15:22 – 00:30:18:21
So do we have
00:30:18:21 – 00:30:21:05
technological availability?
00:30:21:05 – 00:30:23:00
Do we have representation
00:30:23:00 – 00:30:25:16
and understanding and creation?
00:30:25:16 – 00:30:28:09
That’s one part of readiness.
00:30:30:12 – 00:30:33:12
Next slide.
00:30:37:24 – 00:30:39:22
So now we’ll see that there’s
00:30:39:22 – 00:30:41:12
there’s various components here.
00:30:41:12 – 00:30:42:11
So you can see here
00:30:42:11 – 00:30:43:19
where it talks
00:30:43:19 – 00:30:44:12
about readiness
00:30:44:12 – 00:30:45:13
when it comes to sign
00:30:45:13 – 00:30:47:12
language recognition.
00:30:47:12 – 00:30:50:24
And we have readiness of American
00:30:50:24 – 00:30:53:01
the American Deaf community.
00:30:53:01 – 00:30:55:07
That’s another component.
00:30:55:07 – 00:30:58:09
Does the deaf community have an in-depth
00:30:58:09 – 00:31:01:23
understanding of the having good quality,
00:31:02:06 – 00:31:05:00
ethical and technological aspects
00:31:05:00 – 00:31:06:17
when it comes to A.I.
00:31:06:17 – 00:31:08:04
by A.I.?
00:31:08:04 – 00:31:11:20
And the last sign of readiness would be
00:31:11:20 – 00:31:12:14
when it comes
00:31:12:14 – 00:31:15:22
to public and private entities,
00:31:16:09 – 00:31:20:08
Are we ready for this kind of technology
00:31:20:15 – 00:31:22:04
and what the responsibilities
00:31:22:04 – 00:31:24:02
be that come along with that
00:31:24:02 – 00:31:27:02
and the accountability?
00:31:28:04 – 00:31:31:04
Next slide, please.
00:31:34:15 – 00:31:36:06
So as you can see here,
00:31:36:06 – 00:31:37:05
one of the components
00:31:37:05 – 00:31:38:24
that we talked about are civil
00:31:38:24 – 00:31:40:13
rights and civil protections.
00:31:41:12 – 00:31:43:04
So what we have
00:31:43:04 – 00:31:45:23
quality control and certification
00:31:45:23 – 00:31:47:13
when it comes to the interpreters
00:31:47:13 – 00:31:49:01
and certification for the A.I.
00:31:49:01 – 00:31:50:17
technology itself.
00:31:50:17 – 00:31:51:06
Also,
00:31:51:06 – 00:31:53:03
we have to consider the impact
00:31:53:03 – 00:31:55:09
that this would have on the culture,
00:31:55:09 – 00:31:56:18
the current culture
00:31:56:18 – 00:31:58:08
and the culture of the future.
00:31:58:08 – 00:31:59:20
And what
00:31:59:20 – 00:32:00:19
what does
00:32:00:19 – 00:32:03:11
state of the art technology mean?
00:32:03:11 – 00:32:06:05
How would we make sure that we are up
00:32:06:05 – 00:32:07:12
to date with technology
00:32:07:12 – 00:32:09:04
as it changes over time
00:32:09:04 – 00:32:10:02
and ensure
00:32:10:02 – 00:32:11:12
that we have the appropriate
00:32:11:12 – 00:32:13:03
response to those changes?
00:32:13:03 – 00:32:16:03
Next slide.
00:32:26:00 – 00:32:27:08
So I believe that, well,
00:32:27:08 – 00:32:28:22
I can go ahead and do this part.
00:32:28:22 – 00:32:33:06
And so in just a few hours.
00:32:33:08 – 00:32:36:08
Next slide, please.
00:32:40:04 – 00:32:40:21
Okay, great.
00:32:40:21 – 00:32:42:01
Sorry about that.
00:32:42:01 – 00:32:45:22
And okay, So within a few hours,
00:32:46:05 – 00:32:48:08
we had these internal discussions
00:32:48:08 – 00:32:50:00
at the three webinars,
00:32:50:00 – 00:32:51:11
and within those hours
00:32:51:11 – 00:32:52:23
we had a lot of ethical issues
00:32:52:23 – 00:32:54:10
that came up to be discussed
00:32:54:10 – 00:32:56:06
and we touched on ethics,
00:32:56:06 – 00:32:58:10
we touched on fairness and safety.
00:32:58:10 – 00:32:59:09
And I believe
00:32:59:09 – 00:33:00:23
all of these things were covered.
00:33:00:23 – 00:33:02:20
Also, the deaf participants
00:33:02:20 – 00:33:04:21
were able to explain
00:33:04:21 – 00:33:07:08
how we can use our principles
00:33:07:08 – 00:33:09:03
and how we can use our models
00:33:09:03 – 00:33:10:15
in a way that’s ethical
00:33:10:15 – 00:33:11:19
in order to prevent
00:33:11:19 – 00:33:13:00
any ongoing
00:33:13:00 – 00:33:14:00
harm
00:33:14:00 – 00:33:15:09
or anything that could come
00:33:15:09 – 00:33:16:24
against the deaf community.
00:33:16:24 – 00:33:18:00
So we talked about that
00:33:18:00 – 00:33:19:04
fairness and equity
00:33:19:04 – 00:33:20:21
in these conversations,
00:33:20:21 – 00:33:23:09
how it pertains to society.
00:33:23:09 – 00:33:24:00
Next slide.
00:33:29:06 – 00:33:31:18
So we do have a few suggestions
00:33:31:18 – 00:33:35:02
and we suggest that there be a
00:33:35:08 – 00:33:36:23
the appropriate level of protection
00:33:36:23 – 00:33:38:19
and privacy and confidentiality
00:33:38:19 – 00:33:39:20
when it comes to A.I.
00:33:39:20 – 00:33:41:07
backed by A.I.,
00:33:41:07 – 00:33:42:11
and that will allow us
00:33:42:11 – 00:33:43:22
to have better protections
00:33:43:22 – 00:33:44:22
across the Internet
00:33:44:22 – 00:33:47:09
for all kinds of applications.
00:33:47:09 – 00:33:51:05
Also, another suggestion was that we have
00:33:52:15 – 00:33:55:19
that these concerns regarding risk
00:33:55:23 – 00:33:58:21
be established and considered in order
00:33:58:21 – 00:33:59:13
to make sure
00:33:59:13 – 00:34:01:20
that we have strict regulations in place
00:34:01:20 – 00:34:03:21
to avoid any adverse
00:34:03:21 – 00:34:05:10
downstream consequences
00:34:05:10 – 00:34:06:18
for governance, business
00:34:06:18 – 00:34:09:18
and social infrastructures.
00:34:10:13 – 00:34:13:13
Next slide.
00:34:14:15 – 00:34:15:08
Okay.
00:34:15:08 – 00:34:17:06
So this word stress
00:34:17:06 – 00:34:20:06
or socio technical systems
00:34:21:01 – 00:34:23:12
and again asks for short
00:34:23:12 – 00:34:26:07
means that we have to recognize
00:34:26:07 – 00:34:28:07
that we do have technology
00:34:28:07 – 00:34:31:12
and we also have our the socio aspect
00:34:31:20 – 00:34:34:01
when these two things come together,
00:34:34:01 – 00:34:34:19
they’re going to have
00:34:34:19 – 00:34:36:15
an influence on each other.
00:34:36:15 – 00:34:38:03
So we need to always
00:34:38:03 – 00:34:38:20
be sure
00:34:38:20 – 00:34:41:18
to look at the system of technology
00:34:41:18 – 00:34:44:04
while considering its impact on society
00:34:44:04 – 00:34:45:14
and vice versa.
00:34:45:14 – 00:34:47:03
We have to look at how both of these
00:34:47:03 – 00:34:48:20
things interact. Next slide.
00:34:54:12 – 00:34:55:17
So I’ll give you a moment
00:34:55:17 – 00:34:58:17
to read through the slide.
00:35:06:07 – 00:35:07:06
So it’s important.
00:35:07:06 – 00:35:08:09
Here to take note of
00:35:08:09 – 00:35:09:20
is that Estes
00:35:09:20 – 00:35:13:18
refers to how things are correlated, how
00:35:13:18 – 00:35:15:11
these co influences
00:35:15:11 – 00:35:18:01
happen in both society and technology
00:35:18:01 – 00:35:21:01
as it pertains to the organization.
00:35:21:08 – 00:35:23:03
So it’s so important
00:35:23:03 – 00:35:25:04
that during the design process
00:35:25:04 – 00:35:28:03
we consider both of these areas
00:35:28:03 – 00:35:29:19
and look at our results
00:35:29:19 – 00:35:34:04
to ensure that air by air is optimizing
00:35:34:09 – 00:35:37:09
both of these two subsystems.
00:35:41:18 – 00:35:44:19
So the key here is to attend to
00:35:44:19 – 00:35:47:21
how the social behaviors of humans
00:35:47:21 – 00:35:50:00
are going to combine and influence
00:35:50:00 – 00:35:52:16
by the structures of technology
00:35:52:16 – 00:35:55:16
and vice versa.
00:35:56:06 – 00:35:59:06
So these are not two separate things.
00:35:59:15 – 00:36:02:10
We do have a few recommendations
00:36:02:10 – 00:36:04:02
that we would like to share.
00:36:04:02 – 00:36:05:19
The first one is that
00:36:05:19 – 00:36:09:10
it needs to be understood that AI by AI
00:36:09:10 – 00:36:12:11
is a socio tech and technological system
00:36:13:10 – 00:36:15:11
and also we need to
00:36:15:11 – 00:36:16:07
make sure
00:36:16:07 – 00:36:19:10
that the deaf wisdom from the community
00:36:19:15 – 00:36:21:02
is a vital part of this.
00:36:21:02 – 00:36:22:10
We want to use our wisdom
00:36:22:10 – 00:36:24:03
and our experience
00:36:24:03 – 00:36:26:09
from interpreting experiences
00:36:26:09 – 00:36:29:19
to our experience with VR s VR AI,
00:36:29:24 – 00:36:31:01
and also
00:36:31:01 – 00:36:33:09
how we as a community
00:36:33:09 – 00:36:35:08
experience technology.
00:36:35:08 – 00:36:38:21
And we want to be able to impart
00:36:38:24 – 00:36:40:00
our experience
00:36:40:00 – 00:36:43:01
and our wisdom as the framework is being
00:36:43:01 – 00:36:46:01
developed in this realm.
00:36:46:01 – 00:36:50:00
Thirdly, we want to continue to engage
00:36:50:00 – 00:36:51:14
with the deaf community
00:36:51:14 – 00:36:52:21
and build our knowledge
00:36:52:21 – 00:36:55:17
and our awareness through funding.
00:36:55:17 – 00:36:58:17
One example of that funding are grants,
00:36:58:20 – 00:37:00:07
and so an example
00:37:00:07 – 00:37:02:15
is the Civic Innovation grant,
00:37:02:15 – 00:37:04:05
where we can receive money
00:37:04:05 – 00:37:04:21
for this
00:37:04:21 – 00:37:06:02
and continue
00:37:06:02 – 00:37:07:20
to plug in to the deaf community
00:37:07:20 – 00:37:09:04
and get their input
00:37:09:04 – 00:37:10:09
and find what the deaf
00:37:10:09 – 00:37:11:15
community would like to see
00:37:11:15 – 00:37:12:14
and what they feel needs
00:37:12:14 – 00:37:14:04
to be brought to attention.
00:37:14:04 – 00:37:15:08
And as these laws and
00:37:15:08 – 00:37:16:13
policies are developed.
00:37:18:11 – 00:37:21:11
Next slide, please.
00:37:24:21 – 00:37:25:08
And as we
00:37:25:08 – 00:37:26:09
get close to wrapping up,
00:37:26:09 – 00:37:28:13
I’d like to make the point that I by
00:37:28:13 – 00:37:30:04
I will probably never be able
00:37:30:04 – 00:37:31:22
to totally replace
00:37:31:22 – 00:37:32:16
humans
00:37:32:16 – 00:37:35:16
because of the interactivity issues.
00:37:35:20 – 00:37:37:01
Often we can predict
00:37:37:01 – 00:37:39:00
there would be misunderstandings
00:37:39:00 – 00:37:42:00
and so we need to have that deep
00:37:42:05 – 00:37:44:03
human knowledge there.
00:37:44:03 – 00:37:47:03
So we’ll have to look for
00:37:47:15 – 00:37:50:00
a human in the loop design
00:37:50:00 – 00:37:51:06
as this technology
00:37:51:06 – 00:37:53:18
is developed over time.
00:37:53:18 – 00:37:57:03
And in terms of human in the loop design
00:37:57:05 – 00:38:00:05
to explain a bit about about that,
00:38:00:12 – 00:38:01:22
of course we would have the A.I.
00:38:01:22 – 00:38:05:03
technology and often
00:38:05:03 – 00:38:08:03
we see that I, I get information
00:38:08:08 – 00:38:10:05
through different sources,
00:38:10:05 – 00:38:12:14
but we want to be able to emphasize
00:38:12:14 – 00:38:14:12
the importance of A.I.
00:38:14:12 – 00:38:16:00
getting feedback
00:38:16:00 – 00:38:19:23
from human interactivity.
00:38:20:12 – 00:38:22:08
There needs to be a feedback loop
00:38:22:08 – 00:38:23:07
food feedback loop
00:38:23:07 – 00:38:24:05
where humans are always
00:38:24:05 – 00:38:27:12
involved in this process so that
00:38:28:05 – 00:38:30:05
the design and the development
00:38:30:05 – 00:38:31:24
and the operation
00:38:31:24 – 00:38:33:22
and everything is verified
00:38:33:22 – 00:38:34:21
by humans
00:38:34:21 – 00:38:37:02
who are involved with the project.
00:38:37:02 – 00:38:38:17
That would be the end goal.
00:38:38:17 – 00:38:39:18
Next slide, please.
00:38:45:23 – 00:38:46:17
Hello everyone.
00:38:46:17 – 00:38:50:03
I’m back so I would like to
00:38:50:05 – 00:38:53:22
we have the advisory group on AI
00:38:53:22 – 00:38:55:02
and final language interpreting
00:38:55:02 – 00:38:56:00
and this has been
00:38:56:00 – 00:38:57:24
there’s been so many
00:38:57:24 – 00:38:59:06
benefits to this group,
00:38:59:06 – 00:39:00:15
so much AB many hours
00:39:00:15 – 00:39:02:18
and work has gone into this.
00:39:02:18 – 00:39:05:15
And so this isn’t the end of our work.
00:39:05:15 – 00:39:07:00
This is just the springboard
00:39:07:00 – 00:39:08:15
to the future for more discussions
00:39:08:15 – 00:39:09:11
and more partnerships
00:39:09:11 – 00:39:11:00
with the community at large.
00:39:11:00 – 00:39:13:05
So we have an event to share with you.
00:39:13:05 – 00:39:14:21
A Save the Date.
00:39:14:21 – 00:39:17:21
We have a symposium
00:39:18:02 – 00:39:21:02
on AI and sign language interpreting
00:39:21:14 – 00:39:23:10
and it will be hosted on April
00:39:23:10 – 00:39:27:08
20th and 21st is a Saturday and Sunday
00:39:27:08 – 00:39:28:11
this year
00:39:28:11 – 00:39:31:11
and will be here at Brown University.
00:39:32:21 – 00:39:35:04
And we also will have
00:39:35:04 – 00:39:37:13
accessibility options to join us
00:39:37:13 – 00:39:38:14
through Zoom.
00:39:38:14 – 00:39:40:16
Anyone can join from anywhere
00:39:40:16 – 00:39:43:09
we are planning to on Saturday.
00:39:43:09 – 00:39:46:16
It will be from 9 to 6 and Sunday 9
00:39:46:16 – 00:39:49:16
to 2, just a half day on Sunday.
00:39:49:22 – 00:39:52:00
And this is Eastern Standard Time.
00:39:52:24 – 00:39:53:21
This is very
00:39:53:21 – 00:39:55:02
important for us
00:39:55:02 – 00:39:57:20
to bring in different perspectives,
00:39:57:20 – 00:40:00:08
different experts and different people
00:40:00:08 – 00:40:02:16
who have in-depth experience in A.I.
00:40:02:16 – 00:40:04:10
and sign language interpreting
00:40:04:10 – 00:40:05:07
so we can really have
00:40:05:07 – 00:40:06:15
an in-depth discussion
00:40:06:15 – 00:40:08:02
on what this looks like.
00:40:08:02 – 00:40:09:15
And it’s also an opportunity
00:40:09:15 – 00:40:11:12
for us to do a deeper dive
00:40:11:12 – 00:40:12:21
into what our presenters
00:40:12:21 – 00:40:13:13
have really talked
00:40:13:13 – 00:40:14:21
about during this session
00:40:14:21 – 00:40:15:18
as well,
00:40:15:18 – 00:40:17:21
to flesh out different topics and issues
00:40:17:21 – 00:40:18:20
that may arise.
00:40:18:20 – 00:40:20:01
I to see you there.
00:40:20:01 – 00:40:23:01
I’m very excited about it.
00:40:23:11 – 00:40:24:15
I’m going to go ahead
00:40:24:15 – 00:40:27:15
and turn this over to Anne Marie.
00:40:28:03 – 00:40:30:13
Thank you so much, Tim.
00:40:30:13 – 00:40:32:20
So I would just like to add
00:40:32:20 – 00:40:34:11
a little bit more information
00:40:34:11 – 00:40:35:21
for your awareness.
00:40:35:21 – 00:40:38:21
And in terms of the participations,
00:40:39:03 – 00:40:42:08
the participants, we did have 300
00:40:42:20 – 00:40:44:21
people come in that participated
00:40:44:21 – 00:40:46:11
during these webinars.
00:40:46:11 – 00:40:48:15
So we had a very great showing for that
00:40:48:15 – 00:40:50:12
and we were on Zoom,
00:40:50:12 – 00:40:52:05
so we were able to see
00:40:52:05 – 00:40:54:04
a lot of comments in the Q&A.
00:40:54:04 – 00:40:55:21
We saw lots of questions come through,
00:40:55:21 – 00:40:57:05
which was great.
00:40:57:05 – 00:41:00:09
And so after this meeting today,
00:41:00:11 – 00:41:01:15
we can share
00:41:01:15 – 00:41:02:16
some more information with you.
00:41:02:16 – 00:41:04:02
But we had a lot of people
00:41:04:02 – 00:41:06:17
that were coming in
00:41:06:17 – 00:41:09:08
for this research, for these webinars,
00:41:11:08 – 00:41:14:17
and we had eight we had 98%
00:41:14:17 – 00:41:16:12
of the people who participated
00:41:16:12 – 00:41:18:05
sign and agreed to be able
00:41:18:05 – 00:41:20:01
to share their information.
00:41:20:01 – 00:41:25:10
And we also had ASL involved
00:41:25:10 – 00:41:26:24
for all of those participant
00:41:26:24 – 00:41:29:24
participants out of 55,
00:41:31:01 – 00:41:32:00
out of the fit
00:41:32:00 – 00:41:32:20
out of the people
00:41:32:20 – 00:41:35:20
that were on the forum out of,
00:41:36:06 – 00:41:38:18
we had 98% of the people
00:41:38:18 – 00:41:40:07
who were involved agreed
00:41:40:07 – 00:41:43:07
to share the information
00:41:47:00 – 00:41:50:00
and to add to that
00:41:51:05 – 00:41:54:05
in terms of the topics and
00:41:54:15 – 00:41:56:15
what we what we were talking about,
00:41:56:15 – 00:41:58:22
we talked about the values and these
00:41:58:22 – 00:42:01:22
and the perspectives of the community.
00:42:03:15 – 00:42:04:19
We talked about research
00:42:04:19 – 00:42:05:23
findings
00:42:05:23 – 00:42:07:16
and the level of participation
00:42:07:16 – 00:42:08:10
was just great.
00:42:08:10 – 00:42:10:23
We were able to do a lot of research
00:42:10:23 – 00:42:13:13
and go through those topics
00:42:13:13 – 00:42:14:15
and get the specifics
00:42:14:15 – 00:42:15:20
from the participants
00:42:15:20 – 00:42:18:24
so that we had very clear results.
00:42:19:07 – 00:42:20:11
Next slide, please.
00:42:28:01 – 00:42:29:21
And in terms of the
00:42:29:21 – 00:42:31:16
discussion we were on
00:42:31:16 – 00:42:34:03
for about 173 minutes
00:42:34:03 – 00:42:37:10
and we were able to show
00:42:37:11 – 00:42:39:07
a lot of in-depth discussion
00:42:39:07 – 00:42:41:11
and comments that came through.
00:42:41:11 – 00:42:43:22
It gave a lot of value
00:42:43:22 – 00:42:44:21
to our discussion.
00:42:44:21 – 00:42:47:00
We had quite a bit of participation
00:42:47:00 – 00:42:48:18
and consideration to go over
00:42:48:18 – 00:42:50:04
and some of the topics
00:42:50:04 – 00:42:52:12
that were discussed really focused.
00:42:52:12 – 00:42:55:10
We talked about some of those earlier.
00:42:55:10 – 00:42:56:01
For example,
00:42:56:01 – 00:42:57:04
we talked about deaf
00:42:57:04 – 00:42:59:09
community readiness for AI.
00:42:59:09 – 00:43:00:14
Is the deaf community
00:43:00:14 – 00:43:02:04
actually ready for this?
00:43:02:04 – 00:43:06:09
So these discussions were so important
00:43:06:09 – 00:43:08:01
and I think having the participants
00:43:08:01 – 00:43:10:03
there to talk about AI
00:43:10:03 – 00:43:11:06
and to consider
00:43:11:06 – 00:43:11:18
whether or not
00:43:11:18 – 00:43:12:19
they were ready
00:43:12:19 – 00:43:15:21
to have this method of communication
00:43:15:21 – 00:43:19:14
as part of their life, to have AI there.
00:43:20:03 – 00:43:20:23
They discussed whether or
00:43:20:23 – 00:43:21:24
not they were ready for that.
00:43:23:06 – 00:43:24:14
And many of the
00:43:24:14 – 00:43:27:14
participants, 17
00:43:29:17 – 00:43:30:23
many of the participants
00:43:30:23 – 00:43:33:23
that were there from the advisory group,
00:43:37:21 – 00:43:38:14
were involved
00:43:38:14 – 00:43:41:14
in this process.
00:43:43:13 – 00:43:46:13
And
00:43:55:09 – 00:43:56:02
let me go back
00:43:56:02 – 00:43:59:02
for a second.
00:44:08:21 – 00:44:09:03
Okay.
00:44:09:03 – 00:44:12:03
Going back a bit to the process,
00:44:15:23 – 00:44:18:23
the discussion went on for 173 minutes
00:44:19:10 – 00:44:20:13
and we were able
00:44:20:13 – 00:44:22:01
to share different comments.
00:44:22:01 – 00:44:24:18
The participants went over
00:44:24:18 – 00:44:26:08
the things that we talked about today
00:44:26:08 – 00:44:29:08
and also
00:44:30:09 – 00:44:32:18
hashtag Save death.
00:44:32:18 – 00:44:37:06
I became a hashtag that we used
00:44:37:06 – 00:44:40:06
and we got a lot of feedback
00:44:41:21 – 00:44:44:18
and also recommendations.
00:44:44:18 – 00:44:46:16
And we talked about the broader influence
00:44:46:16 – 00:44:49:16
that this will have on the community.
00:44:55:22 – 00:44:56:09
We talked
00:44:56:09 – 00:44:59:09
about the organizations,
00:45:00:19 – 00:45:02:20
and these discussions were very important
00:45:02:20 – 00:45:05:06
to talk about collecting the data
00:45:05:06 – 00:45:08:06
and the process
00:45:09:07 – 00:45:10:01
of.
00:45:10:01 – 00:45:11:06
Next slide, please.
00:45:31:09 – 00:45:32:02
Okay.
00:45:32:02 – 00:45:33:17
So we had three
00:45:33:17 – 00:45:34:15
categories
00:45:34:15 – 00:45:36:02
to flesh out
00:45:36:02 – 00:45:38:00
all of this information through.
00:45:38:00 – 00:45:40:04
We wanted to go through
00:45:40:04 – 00:45:44:00
and with our advisory group members and
00:45:44:00 – 00:45:47:20
we decided for more thematic
00:45:47:20 – 00:45:51:06
areas of research as a team had arisen.
00:45:51:16 – 00:45:53:04
All six of them were shown
00:45:53:04 – 00:45:56:04
in the first passcode book,
00:45:56:12 – 00:45:58:04
and now we’re going to add
00:45:58:04 – 00:46:01:10
that to what we’ll send out to all.
00:46:01:17 – 00:46:04:03
And that’s also included in the report.
00:46:04:03 – 00:46:07:14
I think I saw some questions in the Q A
00:46:07:16 – 00:46:10:16
It is available online.
00:46:10:20 – 00:46:12:22
It will be available
00:46:12:22 – 00:46:16:00
as we’re doing an invitation process
00:46:16:00 – 00:46:19:23
to add that to our website.
00:46:30:18 – 00:46:32:05
I
00:46:32:05 – 00:46:33:03
do you want me to
00:46:33:03 – 00:46:36:03
move on to the next slide?
00:46:36:06 – 00:46:37:04
I think we’re at the end.
00:46:37:04 – 00:46:37:13
Okay.
00:47:00:05 – 00:47:03:05
So
00:47:05:07 – 00:47:06:08
I’m saying if the team
00:47:06:08 – 00:47:08:12
could all turn on their cameras,
00:47:08:12 – 00:47:11:12
we’ll start in with the Q&A.
00:47:16:08 – 00:47:16:19
All right.
00:47:16:19 – 00:47:17:20
Are we ready to address
00:47:17:20 – 00:47:20:03
the questions in the Q&A?
00:47:20:03 – 00:47:22:08
Let’s do it, though.
00:47:22:08 – 00:47:23:21
The first thing I’d like to do
00:47:23:21 – 00:47:25:19
is thank you all for your questions.
00:47:25:19 – 00:47:28:19
Thank you so very much.
00:47:29:04 – 00:47:29:24
The first question
00:47:29:24 – 00:47:32:24
that we have for the panel
00:47:34:01 – 00:47:35:16
in the future,
00:47:35:16 – 00:47:38:16
say 10 to 15 years down the road,
00:47:38:20 – 00:47:39:08
A.I.
00:47:39:08 – 00:47:42:08
by A.I., what will that look like?
00:47:43:21 – 00:47:46:21
How many applications
00:47:46:22 – 00:47:48:12
are possible?
00:47:48:12 – 00:47:53:08
For example, TV theaters, movie theaters?
00:47:53:19 – 00:47:55:16
Will there be an automatic interpreter
00:47:55:16 – 00:47:58:16
pop up on the screen?
00:48:00:08 – 00:48:01:12
What do you think
00:48:01:12 – 00:48:02:24
is the potential for A.I.
00:48:02:24 – 00:48:05:24
in the future?
00:48:08:21 – 00:48:10:14
Jeff Would you like to go?
00:48:10:14 – 00:48:13:05
Jeff saying yes, I will talk about that.
00:48:13:05 – 00:48:13:24
I bet
00:48:13:24 – 00:48:15:19
I will have so many applications
00:48:15:19 – 00:48:16:14
in the future.
00:48:16:14 – 00:48:18:07
I think in 10 to 15 years,
00:48:18:07 – 00:48:20:09
the possibilities are astounding.
00:48:20:09 – 00:48:23:23
There’s you know, it I would say that
00:48:24:00 – 00:48:25:08
computes two equates
00:48:25:08 – 00:48:28:00
to about 100 years of human development
00:48:28:00 – 00:48:29:06
because technology moves
00:48:29:06 – 00:48:30:21
at such a rapid pace.
00:48:30:21 – 00:48:31:10
One thing
00:48:31:10 – 00:48:33:03
I would like to say that for sure,
00:48:33:03 – 00:48:34:19
I know that it will have improved
00:48:34:19 – 00:48:36:02
drastically by then
00:48:36:02 – 00:48:38:00
because I think it’s going to continue
00:48:38:00 – 00:48:39:16
marching on
00:48:39:16 – 00:48:41:14
and improving as time goes by.
00:48:41:14 – 00:48:43:12
Now, will we be more trusting
00:48:43:12 – 00:48:44:18
of these applications
00:48:44:18 – 00:48:47:18
in the future and more confident with it?
00:48:47:19 – 00:48:51:10
I think in low risk situations,
00:48:52:22 – 00:48:54:14
I don’t think it should be too bad.
00:48:54:14 – 00:48:57:14
For example,
00:48:57:14 – 00:48:59:18
not a police encounter,
00:48:59:18 – 00:49:00:17
medical encounter
00:49:00:17 – 00:49:03:20
or anything of that sort, but I bet
00:49:03:20 – 00:49:06:12
I could be used for automatic
00:49:06:12 – 00:49:08:09
conversation and dialogs like,
00:49:08:09 – 00:49:10:04
you know, with robocalls,
00:49:10:04 – 00:49:12:14
sharing information,
00:49:12:14 – 00:49:13:24
having something set,
00:49:13:24 – 00:49:17:13
you know, an automated system.
00:49:18:03 – 00:49:19:10
But in the future
00:49:19:10 – 00:49:22:10
I foresee more captioning being involved,
00:49:22:17 – 00:49:24:20
not necessarily ASL only,
00:49:24:20 – 00:49:26:19
but I think captioning
00:49:26:19 – 00:49:29:16
and multilingual captioning as a whole
00:49:29:16 – 00:49:32:06
will have developed so much over time.
00:49:32:06 – 00:49:33:21
I think there are so many possibilities
00:49:33:21 – 00:49:34:16
and different directions
00:49:34:16 – 00:49:35:23
it could go again.
00:49:35:23 – 00:49:37:11
It’s hard to predict everything
00:49:37:11 – 00:49:38:19
that may happen in the future,
00:49:38:19 – 00:49:39:20
but in a nutshell,
00:49:39:20 – 00:49:42:01
i believe that’s what would happen
00:49:42:01 – 00:49:45:03
and I think that some examples of that
00:49:45:08 – 00:49:46:13
do so.
00:49:46:13 – 00:49:47:06
For example,
00:49:47:06 – 00:49:49:21
when you’re driving through a place
00:49:49:21 – 00:49:51:20
and you want to order coffee
00:49:51:20 – 00:49:53:12
or you want to order food
00:49:53:12 – 00:49:55:22
and that kind of situation,
00:49:55:22 – 00:49:58:13
and when you’re ordering your food,
00:49:58:13 – 00:49:59:00
you know
00:50:00:04 – 00:50:00:14
it.
00:50:00:14 – 00:50:01:05
It’s not
00:50:01:05 – 00:50:02:03
going to be something
00:50:02:03 – 00:50:03:10
that’s disastrous to your life
00:50:03:10 – 00:50:05:00
if something messes up.
00:50:05:00 – 00:50:06:20
But when it comes to the barriers,
00:50:06:20 – 00:50:09:01
the deaf community experiences
00:50:09:01 – 00:50:10:12
in other situations,
00:50:10:12 – 00:50:12:06
there’s so many different possibilities
00:50:12:06 – 00:50:13:17
of how this could go.
00:50:13:17 – 00:50:15:04
And I think, again,
00:50:15:04 – 00:50:17:14
we have to make sure that the technology
00:50:17:14 – 00:50:19:08
is ready, verified
00:50:19:08 – 00:50:21:16
and it’s going to be more beneficial
00:50:21:16 – 00:50:23:18
and and not harmful.
00:50:23:18 – 00:50:25:18
I think that’s what we have to see,
00:50:25:18 – 00:50:27:02
you know?
00:50:27:02 – 00:50:27:19
Theresa here,
00:50:27:19 – 00:50:29:03
I’d like to make a comment as well,
00:50:29:03 – 00:50:30:15
and I’d like to emphasize the point
00:50:30:15 – 00:50:34:01
about low risk situations for a moment.
00:50:34:01 – 00:50:35:06
I think that’s important.
00:50:35:06 – 00:50:37:08
It’s imperative for us to realize
00:50:37:08 – 00:50:38:23
that we always have a need
00:50:38:23 – 00:50:41:23
for human interpreters,
00:50:41:24 – 00:50:43:12
specific situations
00:50:43:12 – 00:50:44:24
like medical situations,
00:50:44:24 – 00:50:47:06
legal situations, court
00:50:47:06 – 00:50:49:04
law enforcement interactions.
00:50:49:04 – 00:50:50:07
It’s really imperative
00:50:50:07 – 00:50:51:24
that we include the human aspect
00:50:51:24 – 00:50:54:08
and human judgment in those situations
00:50:55:19 – 00:50:56:24
in this century.
00:50:56:24 – 00:51:00:19
And to add to that, I think one thing
00:51:00:19 – 00:51:01:16
that we also have to
00:51:01:16 – 00:51:04:16
emphasize is knowing the language itself
00:51:04:19 – 00:51:07:06
and knowing that process,
00:51:07:06 – 00:51:09:14
whether the responses
00:51:09:14 – 00:51:11:00
and also looking at
00:51:11:00 – 00:51:12:22
whether it’s a live person,
00:51:12:22 – 00:51:16:06
if it’s going to be a mixed situation,
00:51:16:10 – 00:51:17:15
we have to recognize
00:51:17:15 – 00:51:18:23
exactly what is involved
00:51:18:23 – 00:51:20:21
situation by situation
00:51:20:21 – 00:51:24:08
so that we can make sure that the access
00:51:24:08 – 00:51:25:21
that’s being provided is as close
00:51:25:21 – 00:51:27:17
to 100% accurate as possible.
00:51:27:17 – 00:51:29:16
So like Theresa said, humans
00:51:29:16 – 00:51:31:13
have to be involved in this
00:51:31:13 – 00:51:33:05
and in this feedback loop,
00:51:33:05 – 00:51:35:15
and it’s very critical that we
00:51:35:15 – 00:51:37:00
emphasize that.
00:51:37:00 – 00:51:37:16
Theresa saying,
00:51:37:16 – 00:51:39:01
I’d like to add a little bit more.
00:51:39:01 – 00:51:40:01
One thing to think about
00:51:40:01 – 00:51:41:12
is in my past experience,
00:51:41:12 – 00:51:45:13
for many years, living in Mexico,
00:51:45:13 – 00:51:47:00
the deaf population
00:51:47:00 – 00:51:48:20
in the community, in New Mexico,
00:51:48:20 – 00:51:51:22
New Mexico, you know,
00:51:52:19 – 00:51:56:00
they were able to be
00:51:56:00 – 00:51:57:15
mainstreamed into the school
00:51:59:14 – 00:52:00:20
and the dialects are
00:52:00:20 – 00:52:02:22
different in different areas.
00:52:02:22 – 00:52:05:06
I may not recognize that.
00:52:05:06 – 00:52:06:14
So growing up in those schools
00:52:06:14 – 00:52:08:13
where I where the communities were small,
00:52:08:13 – 00:52:10:07
the dialects were diverse,
00:52:10:07 – 00:52:12:21
I may not have that capability
00:52:12:21 – 00:52:15:05
just yet to recognize that.
00:52:15:05 – 00:52:17:17
So we need to make sure that our air
00:52:17:17 – 00:52:20:17
is ready for this specifically,
00:52:20:18 – 00:52:21:14
for example,
00:52:21:14 – 00:52:24:14
with our black deaf signers,
00:52:24:21 – 00:52:26:08
the American sign language
00:52:26:08 – 00:52:27:02
that they use
00:52:27:02 – 00:52:28:17
is so culturally diverse
00:52:28:17 – 00:52:30:12
than American Sign language.
00:52:30:12 – 00:52:32:00
We have to make sure
00:52:32:00 – 00:52:32:19
that we have the right
00:52:32:19 – 00:52:35:10
understanding of interpreting.
00:52:35:10 – 00:52:38:10
And for this approach.
00:52:43:14 – 00:52:44:03
Yes.
00:52:44:03 – 00:52:46:19
And more questions.
00:52:46:19 – 00:52:47:17
Very exciting.
00:52:47:17 – 00:52:49:12
Okay, so another question.
00:52:49:12 – 00:52:52:22
Question number two is related
00:52:52:22 – 00:52:55:22
to the term
00:52:56:06 – 00:52:59:06
sign language
00:53:03:19 – 00:53:05:01
and atomization.
00:53:05:01 – 00:53:08:01
00:53:09:05 – 00:53:09:23
So basically
00:53:09:23 – 00:53:12:23
related to privacy,
00:53:13:03 – 00:53:15:23
protecting individuals and their data.
00:53:18:00 – 00:53:19:14
So from
00:53:19:14 – 00:53:20:21
the research and the report
00:53:20:21 – 00:53:22:15
that you all produced,
00:53:22:15 – 00:53:24:20
can you guys discuss and share a bit
00:53:24:20 – 00:53:27:20
about the data privacy
00:53:28:18 – 00:53:30:06
of confidentiality
00:53:30:06 – 00:53:33:14
and also for automatic interpretation,
00:53:33:23 – 00:53:36:24
would you use an avatar and
00:53:37:10 – 00:53:38:13
what does this look like?
00:53:38:13 – 00:53:39:22
How would you be able to
00:53:39:22 – 00:53:41:06
represent individuals
00:53:41:06 – 00:53:44:06
while also protecting them?
00:53:44:09 – 00:53:44:24
Jeff Here
00:53:44:24 – 00:53:46:03
I would like to take that one,
00:53:46:03 – 00:53:47:18
if that’s all right with the group.
00:53:47:18 – 00:53:50:12
So I prefer to sign
00:53:50:12 – 00:53:51:20
automatic interpreting
00:53:51:20 – 00:53:53:14
for data collection.
00:53:53:14 – 00:53:56:00
The faith is involved in that.
00:53:56:00 – 00:53:58:21
So one aspect of data collection is worth
00:53:58:21 – 00:54:00:05
signing is
00:54:00:05 – 00:54:01:09
they don’t have our faith,
00:54:01:09 – 00:54:02:18
and it’s imperative
00:54:02:18 – 00:54:05:15
to compare that with speech recognition
00:54:05:15 – 00:54:07:06
through automatic speech recognition.
00:54:07:06 – 00:54:08:23
You have voice and intonation,
00:54:08:23 – 00:54:09:15
but with signing,
00:54:09:15 – 00:54:11:06
if you don’t include the faith,
00:54:11:06 – 00:54:12:22
that’s a deep part of the language
00:54:12:22 – 00:54:14:15
that’s missing itself.
00:54:14:15 – 00:54:18:16
So there’s no real easy way to avoid that
00:54:18:16 – 00:54:20:24
with data collection,
00:54:20:24 – 00:54:22:23
because we have to have our faith
00:54:22:23 – 00:54:25:13
for the language foundation and tone
00:54:25:13 – 00:54:27:02
and meaning to be there.
00:54:27:02 – 00:54:28:13
So we have to be very careful
00:54:28:13 – 00:54:30:08
with the data itself.
00:54:30:08 – 00:54:31:24
We have to protect that
00:54:31:24 – 00:54:32:12
to make sure
00:54:32:12 – 00:54:35:03
that people are fully informed.
00:54:35:03 – 00:54:37:10
Now, informed consent,
00:54:37:10 – 00:54:38:22
we’re going to use that in the training,
00:54:38:22 – 00:54:40:02
in AI by AI.
00:54:40:02 – 00:54:41:20
We will be talking about that
00:54:41:20 – 00:54:42:23
facial recognition
00:54:42:23 – 00:54:44:23
and other identifying information.
00:54:44:23 – 00:54:47:07
My background may be there as well.
00:54:47:07 – 00:54:49:22
So we have to really filter out
00:54:49:22 – 00:54:51:04
that information
00:54:51:04 – 00:54:54:03
and see what part of that is protected
00:54:54:03 – 00:54:56:11
and what part of it is stored.
00:54:56:11 – 00:54:58:00
And we also have to think about
00:54:58:00 – 00:55:00:06
what is private.
00:55:00:06 – 00:55:02:21
And with avatars, you know,
00:55:02:21 – 00:55:03:22
it may be possible
00:55:03:22 – 00:55:06:11
to produce signs with avatars,
00:55:06:11 – 00:55:08:03
but it could be a little bit off.
00:55:08:03 – 00:55:11:03
For example, the avatar.
00:55:11:07 – 00:55:13:14
What can we train that avatar
00:55:13:14 – 00:55:14:18
to make the signs?
00:55:14:18 – 00:55:16:24
And can we identify that
00:55:16:24 – 00:55:18:02
and know that that’s
00:55:18:02 – 00:55:20:03
what the person is signing?
00:55:20:03 – 00:55:21:23
That can be a little bit ambiguous
00:55:21:23 – 00:55:23:00
in trying to identify
00:55:23:00 – 00:55:26:15
what the data and the the machine
00:55:26:15 – 00:55:29:22
learning is actually tracking in.
00:55:31:18 – 00:55:34:15
All of that goes back to
00:55:34:15 – 00:55:37:15
the subject of informed consent.
00:55:39:02 – 00:55:40:23
And I’d like to add
00:55:40:23 – 00:55:42:17
to that one more thing.
00:55:42:17 – 00:55:45:18
In terms of informed consent, often
00:55:45:18 – 00:55:47:07
we think, okay,
00:55:47:07 – 00:55:48:20
this is just a one time thing
00:55:48:20 – 00:55:51:08
I’m signing and I’m giving my consent,
00:55:51:08 – 00:55:52:24
but informed consent
00:55:52:24 – 00:55:54:08
really needs to happen
00:55:54:08 – 00:55:56:03
on an ongoing basis.
00:55:56:03 – 00:55:57:17
We need to be reminded
00:55:57:17 – 00:55:58:20
and we need to make sure
00:55:58:20 – 00:56:01:06
that we continue to give that consent
00:56:01:06 – 00:56:04:07
and continue to agree and remind people
00:56:04:13 – 00:56:06:11
that they have permission
00:56:06:11 – 00:56:07:13
to remove themselves
00:56:07:13 – 00:56:09:14
or to take that consent back.
00:56:09:14 – 00:56:09:23
00:56:09:23 – 00:56:10:19
And so they have
00:56:10:19 – 00:56:11:16
the right to be involved.
00:56:11:16 – 00:56:14:16
They had the right to decline.
00:56:19:04 – 00:56:20:09
Okay.
00:56:20:09 – 00:56:21:19
QUESTION
00:56:21:19 – 00:56:23:01
Thank you so much for your question.
00:56:23:01 – 00:56:23:21
So the next question
00:56:23:21 – 00:56:24:21
we’re going to address
00:56:24:21 – 00:56:28:01
is I’d like to add a little bit to
00:56:28:01 – 00:56:29:23
this as well.
00:56:32:08 – 00:56:35:08
The discussion about
00:56:35:23 – 00:56:38:23
I’m sorry, the discussion about how
00:56:40:18 – 00:56:43:24
this task force and this advisory group
00:56:44:05 – 00:56:45:20
has collaborated
00:56:45:20 – 00:56:48:02
with other organizations,
00:56:48:02 – 00:56:50:15
the NAD Gallaudet University,
00:56:50:15 – 00:56:53:18
other educational bodies.
00:56:53:18 – 00:56:55:10
What does that look like?
00:56:55:10 – 00:56:58:10
What does your partnership look like?
00:57:02:05 – 00:57:04:02
And I can answer that question.
00:57:04:02 – 00:57:08:15
And so I think that first of all,
00:57:09:03 – 00:57:09:24
the advisory
00:57:09:24 – 00:57:12:07
council is really a device,
00:57:12:07 – 00:57:13:18
a diverse group.
00:57:13:18 – 00:57:16:14
We welcome people
00:57:16:14 – 00:57:18:15
from various organizations
00:57:18:15 – 00:57:21:06
to make sure that we’re reflecting
00:57:21:06 – 00:57:23:00
various people, like, for example,
00:57:23:00 – 00:57:24:22
from Gallaudet University,
00:57:24:22 – 00:57:28:16
people from Nairobi, people from Nadi.
00:57:29:00 – 00:57:32:00
We have participants from all over, and
00:57:32:06 – 00:57:33:18
I think that’s important.
00:57:33:18 – 00:57:36:01
But also we want to recognize
00:57:36:01 – 00:57:37:06
that we want to continue
00:57:37:06 – 00:57:40:22
to involve individuals like, for example,
00:57:40:22 – 00:57:45:10
management.
00:57:45:10 – 00:57:47:03
Omar That’s another one.
00:57:47:03 – 00:57:51:12
We want to have diverse representation
00:57:51:12 – 00:57:52:21
to be able to continue
00:57:52:21 – 00:57:54:14
to working with everyone,
00:57:54:14 – 00:57:56:22
because there are groups that have been
00:57:56:22 – 00:57:59:15
have not been included
00:57:59:15 – 00:58:01:05
in these types of processes in the past.
00:58:01:05 – 00:58:03:12
So we want to make sure that our research
00:58:03:12 – 00:58:05:06
and our study can continue
00:58:05:06 – 00:58:06:17
that kind of collaboration
00:58:06:17 – 00:58:08:06
that we’ve already established.
00:58:08:06 – 00:58:08:22
Like I said,
00:58:08:22 – 00:58:12:07
whether it be with Gallaudet University,
00:58:12:07 – 00:58:13:09
other organizations, I’m
00:58:13:09 – 00:58:14:04
not sure if anyone else
00:58:14:04 – 00:58:15:20
has something to add to that comment.
00:58:17:23 – 00:58:19:07
I’d like to add to the comment.
00:58:19:07 – 00:58:20:15
So we have individuals
00:58:20:15 – 00:58:23:15
with specific skills and backgrounds.
00:58:23:15 – 00:58:27:08
We have work where, you know, I myself
00:58:27:08 – 00:58:28:19
work at Gallaudet University.
00:58:28:19 – 00:58:32:01
I’m a research focused on bioethics
00:58:32:01 – 00:58:35:12
and I volunteer to participate
00:58:35:12 – 00:58:37:09
as an individual researcher.
00:58:37:09 – 00:58:38:15
I’m not representing
00:58:38:15 – 00:58:39:14
Gallaudet University,
00:58:39:14 – 00:58:40:13
but at the same time,
00:58:40:13 – 00:58:41:24
we do have discussions
00:58:41:24 – 00:58:44:24
with many people in Gallaudet University
00:58:45:09 – 00:58:47:03
throughout different areas,
00:58:47:03 – 00:58:48:23
recognizing the importance of this.
00:58:48:23 – 00:58:50:01
And sometimes we have a person
00:58:50:01 – 00:58:51:08
who’s willing to work
00:58:51:08 – 00:58:53:06
who is working for Gallaudet University,
00:58:53:06 – 00:58:54:10
but they’re not representing
00:58:54:10 – 00:58:55:24
the university itself
00:58:55:24 – 00:58:58:07
in their role in this research.
00:58:58:07 – 00:59:00:14
So in my opinion, my research
00:59:00:14 – 00:59:01:21
and what I’m looking at
00:59:01:21 – 00:59:03:19
is not representing the university
00:59:03:19 – 00:59:05:20
at large itself.
00:59:05:20 – 00:59:07:18
So I hope that clarifies that
00:59:07:18 – 00:59:08:14
to a degree.
00:59:11:14 – 00:59:13:19
This is Anne-Marie and
00:59:13:19 – 00:59:17:09
I think in terms of education
00:59:17:09 – 00:59:20:11
and nonprofit groups and organizations,
00:59:20:11 – 00:59:21:22
as an ADDY
00:59:21:22 – 00:59:23:00
and other organizations
00:59:23:00 – 00:59:24:14
that we’ve had involved,
00:59:24:14 – 00:59:26:15
these different signing
00:59:26:15 – 00:59:29:03
groups are working closely together,
00:59:29:03 – 00:59:32:03
different companies who have technology.
00:59:32:08 – 00:59:33:19
And I’d like to emphasize
00:59:33:19 – 00:59:36:22
that we are all working together
00:59:37:06 – 00:59:40:22
and that the thing is to have
00:59:41:03 – 00:59:42:14
a platform for discussion
00:59:42:14 – 00:59:44:00
for the deaf community.
00:59:44:00 – 00:59:45:08
And I know often
00:59:45:08 – 00:59:46:07
a lot of individuals
00:59:46:07 – 00:59:47:21
are overlooked in our community
00:59:47:21 – 00:59:48:15
are they’re pushed
00:59:48:15 – 00:59:50:15
out of these discussions.
00:59:50:15 – 00:59:51:16
They don’t get the opportunity
00:59:51:16 – 00:59:53:12
to explain their perspective.
00:59:53:12 – 00:59:55:04
There’s education issues.
00:59:55:04 – 00:59:59:20
And so, you know, I think that A.I.
01:00:00:00 – 01:00:02:06
here and we often see people say,
01:00:02:06 – 01:00:03:23
you know, AI is here.
01:00:03:23 – 01:00:05:13
What is that going to look like?
01:00:05:13 – 01:00:06:23
And some people say, no,
01:00:06:23 – 01:00:08:03
we don’t want to see this happen.
01:00:08:03 – 01:00:09:09
We don’t want A.I.
01:00:09:09 – 01:00:11:21
to part of this interpreting process
01:00:11:21 – 01:00:14:11
while we know it’s coming.
01:00:14:11 – 01:00:16:11
And so I think it’s our responsibility
01:00:16:11 – 01:00:18:21
to ensure that this collaborative effort
01:00:18:21 – 01:00:20:17
stays in place
01:00:20:17 – 01:00:21:24
so that we have deaf community
01:00:21:24 – 01:00:23:18
representation and representation
01:00:23:18 – 01:00:26:00
in the development of AI over time,
01:00:26:00 – 01:00:27:16
not just for education,
01:00:27:16 – 01:00:30:16
but also that these all over America,
01:00:30:21 – 01:00:32:04
they have the responsibility
01:00:32:04 – 01:00:34:08
of ensuring that deaf individuals are
01:00:34:08 – 01:00:35:07
they’re at the table
01:00:35:07 – 01:00:36:14
when these discussions happen.
01:00:40:00 – 01:00:42:16
Let’s move on to the next question.
01:00:42:16 – 01:00:42:23
Okay.
01:00:42:23 – 01:00:44:21
So for our next question,
01:00:44:21 – 01:00:47:14
what are we working on related
01:00:47:14 – 01:00:50:20
to the legislation
01:00:50:20 – 01:00:53:20
and for protection of the deaf community?
01:00:54:07 – 01:00:57:06
What is what does that look like?
01:00:57:06 – 01:00:58:18
I could take that.
01:00:58:18 – 01:01:01:19
So in terms of this, we
01:01:02:22 – 01:01:04:11
I can maybe probably clarify
01:01:04:11 – 01:01:05:13
a bit of the explanation
01:01:05:13 – 01:01:07:07
we’ve given already, but
01:01:07:07 – 01:01:08:24
when it comes to structure,
01:01:08:24 – 01:01:11:06
we do have our advisory group.
01:01:11:06 – 01:01:14:06
We also have safe A.I..
01:01:14:06 – 01:01:16:00
It’s a task force.
01:01:16:00 – 01:01:18:11
And that task force represents
01:01:18:11 – 01:01:20:12
so many different languages
01:01:20:12 – 01:01:23:12
and interpreters, providers
01:01:23:21 – 01:01:25:19
and tech people
01:01:25:19 – 01:01:27:21
that are involved in the tech development
01:01:27:21 – 01:01:28:22
of it.
01:01:28:22 – 01:01:31:00
We also have consumers
01:01:31:00 – 01:01:32:21
who would be using the services,
01:01:32:21 – 01:01:33:20
so we have a broad
01:01:33:20 – 01:01:34:20
range of people involved
01:01:34:20 – 01:01:36:05
in the task force.
01:01:36:05 – 01:01:39:24
Now, in terms of the deaf perspective,
01:01:39:24 – 01:01:43:03
we wanted to make sure that
01:01:43:19 – 01:01:44:19
they got involved,
01:01:44:19 – 01:01:46:18
but it was not until a bit later
01:01:46:18 – 01:01:47:12
after task
01:01:47:12 – 01:01:49:00
Force was formed that they got involved.
01:01:49:00 – 01:01:49:14
And so we said,
01:01:49:14 – 01:01:50:08
you know, we want silos,
01:01:50:08 – 01:01:51:00
which include we want
01:01:51:00 – 01:01:52:08
all of this included.
01:01:52:08 – 01:01:55:08
So we established the advisory council
01:01:55:08 – 01:01:58:04
to ensure that AI and say sign language
01:01:58:04 – 01:01:59:23
interpreting was represented
01:01:59:23 – 01:02:01:06
from the deaf perspective.
01:02:01:06 – 01:02:02:08
And of course
01:02:02:08 – 01:02:03:18
there’s diverse organizations,
01:02:03:18 – 01:02:05:15
the academic perspective,
01:02:05:15 – 01:02:07:10
designers, developers, all of that.
01:02:07:10 – 01:02:10:18
They were involved and in the group
01:02:11:02 – 01:02:11:11
so that
01:02:11:11 – 01:02:14:18
they could give their advice to say, I
01:02:15:06 – 01:02:17:03
now say saved by
01:02:17:03 – 01:02:18:02
AI has been working
01:02:18:02 – 01:02:19:08
with the advisory group
01:02:19:08 – 01:02:20:22
and the goal is to continue
01:02:20:22 – 01:02:24:19
to develop the policies and the law.
01:02:24:24 – 01:02:27:21
Suggestions that we have for the group
01:02:27:21 – 01:02:31:19
so that our parts and our recommendations
01:02:31:19 – 01:02:32:12
are included
01:02:32:12 – 01:02:33:20
in their reporting
01:02:33:20 – 01:02:36:08
and in their data collection and surveys.
01:02:36:08 – 01:02:37:07
Right now
01:02:37:07 – 01:02:38:18
there are ten different languages
01:02:38:18 – 01:02:39:23
that are being looked at.
01:02:39:23 – 01:02:41:06
And so in terms of American
01:02:41:06 – 01:02:44:15
Sign Language, it was noted
01:02:44:15 – 01:02:46:04
that there
01:02:46:04 – 01:02:47:23
needed to be another opportunity
01:02:47:23 – 01:02:48:14
to collaborate
01:02:48:14 – 01:02:50:02
more with the deaf community
01:02:50:02 – 01:02:51:05
to send out surveys
01:02:51:05 – 01:02:52:10
in American Sign Language.
01:02:52:10 – 01:02:53:12
So that’s where we are right now.
01:02:53:12 – 01:02:54:13
In the process.
01:02:54:13 – 01:02:56:17
We’re hoping that our dialog
01:02:56:17 – 01:02:57:17
and our discussion
01:02:57:17 – 01:03:00:04
and our collection of information
01:03:00:04 – 01:03:03:17
will become a great contributor
01:03:03:17 – 01:03:05:08
to the bigger picture.
01:03:05:08 – 01:03:06:23
Does anyone have anything to add?
01:03:06:23 – 01:03:08:12
I just wanted to clarify a comment
01:03:08:12 – 01:03:09:14
that I had made earlier.
01:03:15:22 – 01:03:16:19
Okay.
01:03:16:19 – 01:03:19:19
Moving on to the next question
01:03:20:02 – 01:03:23:03
for this report and research.
01:03:23:23 – 01:03:26:06
We are focusing on ASL.
01:03:26:06 – 01:03:28:10
What about other signed languages
01:03:28:10 – 01:03:29:15
in other countries?
01:03:29:15 – 01:03:32:03
Will we be looking at others
01:03:32:03 – 01:03:35:03
in the future?
01:03:36:13 – 01:03:37:22
I can take that question.
01:03:37:22 – 01:03:39:12
It would be great.
01:03:39:12 – 01:03:40:16
It’s a dream
01:03:40:16 – 01:03:43:10
not for ASL to be the only language
01:03:43:10 – 01:03:44:06
that we look at, right?
01:03:44:06 – 01:03:46:04
We want to consider all of this,
01:03:46:04 – 01:03:48:05
but in terms of the task force
01:03:48:05 – 01:03:51:08
and the advisory council where
01:03:51:12 – 01:03:53:01
we’re working with different
01:03:53:01 – 01:03:54:19
American organizations
01:03:54:19 – 01:03:57:20
and we have been mostly focusing on ASL,
01:03:58:01 – 01:04:01:01
but we are seeing more effort right now
01:04:01:01 – 01:04:02:17
in other parts of the world
01:04:02:17 – 01:04:04:05
where they are focusing
01:04:04:05 – 01:04:06:09
on the automatic interpreting
01:04:06:09 – 01:04:08:01
for other languages.
01:04:08:01 – 01:04:11:21
And so I am aware of the fact
01:04:11:21 – 01:04:14:15
that there are some places in Europe
01:04:14:15 – 01:04:15:23
that are focusing on things.
01:04:15:23 – 01:04:17:09
I don’t have the specific names.
01:04:17:09 – 01:04:19:19
They don’t they don’t come to me.
01:04:19:19 – 01:04:20:21
These names are not coming to me
01:04:20:21 – 01:04:21:06
right now.
01:04:21:06 – 01:04:25:06
But yeah, and Tim can talk more about
01:04:26:05 – 01:04:26:12
one of
01:04:26:12 – 01:04:29:12
those organizations.
01:04:31:16 – 01:04:32:16
We do have people involved
01:04:32:16 – 01:04:33:23
in Europe and Canada.
01:04:33:23 – 01:04:35:10
I know that there are many more
01:04:35:10 – 01:04:36:09
all over the world
01:04:36:09 – 01:04:37:09
who are also looking
01:04:37:09 – 01:04:39:05
at the same technology,
01:04:39:05 – 01:04:42:20
but because right now, Safe Eye
01:04:42:21 – 01:04:45:01
Task Force is focusing on
01:04:45:01 – 01:04:47:04
American policy legislation
01:04:47:04 – 01:04:50:12
and all of that current focus is
01:04:50:12 – 01:04:52:20
specifically what’s happening in America
01:04:52:20 – 01:04:53:21
and North America.
01:04:53:21 – 01:04:55:01
But at the same time,
01:04:55:01 – 01:04:58:13
we could have an impact on Canada
01:04:58:22 – 01:04:59:18
some of our research,
01:04:59:18 – 01:05:01:11
could impact Europe as well.
01:05:01:11 – 01:05:03:17
So I think this is with the process,
01:05:03:17 – 01:05:06:17
we will probably
01:05:07:02 – 01:05:09:01
continue to move forward
01:05:09:01 – 01:05:12:01
and see more replication
01:05:12:01 – 01:05:15:01
of our studies or expansion of our focus.
01:05:15:06 – 01:05:17:05
Looking at it on a more global scale
01:05:17:05 – 01:05:19:08
as we move forward.
01:05:19:08 – 01:05:21:24
Theresa Here I’d like to also add as well
01:05:21:24 – 01:05:23:01
that I think it’s imperative
01:05:23:01 – 01:05:24:12
that we recognize people
01:05:24:12 – 01:05:26:17
using sign language here in the U.S.
01:05:26:17 – 01:05:29:02
are not only using ASL.
01:05:29:02 – 01:05:30:19
So it’s important for us to know
01:05:30:19 – 01:05:32:23
that there are many foreign
01:05:32:23 – 01:05:35:16
languages used here in the US
01:05:35:16 – 01:05:38:08
and we’ve feedback from users of sign
01:05:38:08 – 01:05:39:09
language in the U.S.,
01:05:39:09 – 01:05:41:15
not just American Sign language.
01:05:41:15 – 01:05:45:06
We will or we will be reaching out to
01:05:45:07 – 01:05:46:11
other sign languages
01:05:46:11 – 01:05:48:15
and pulling in that data and information
01:05:48:15 – 01:05:50:11
and their experiences as well.
01:05:50:11 – 01:05:51:08
And it’s important
01:05:51:08 – 01:05:54:14
for us to include everyone at the table.
01:05:58:22 – 01:05:59:12
Okay.
01:05:59:12 – 01:06:01:01
Next question.
01:06:01:01 – 01:06:04:01
So this question is related to readiness.
01:06:05:21 – 01:06:07:20
So in terms
01:06:07:20 – 01:06:10:04
of these technological companies
01:06:10:04 – 01:06:11:12
in development
01:06:11:12 – 01:06:13:16
and a lot of these people
01:06:13:16 – 01:06:14:06
who are working
01:06:14:06 – 01:06:14:24
in this
01:06:14:24 – 01:06:16:10
arena and make these decisions
01:06:16:10 – 01:06:18:08
don’t always include deaf people.
01:06:18:08 – 01:06:21:04
So what approach do we think?
01:06:21:04 – 01:06:23:08
Or how can we make sure
01:06:23:08 – 01:06:24:17
that deaf individuals
01:06:24:17 – 01:06:25:07
are involved
01:06:25:07 – 01:06:26:09
in these conversations
01:06:26:09 – 01:06:27:02
and that they’re always
01:06:27:02 – 01:06:31:07
a part of the process for AI by AI,
01:06:32:06 – 01:06:35:06
machine learning and everything are
01:06:37:02 – 01:06:38:12
Emery Here, I’ll take that.
01:06:38:12 – 01:06:39:01
I’d be happy
01:06:39:01 – 01:06:40:18
to make a comment about that.
01:06:40:18 – 01:06:42:18
So with regards to
01:06:42:18 – 01:06:44:13
what we’ve been speaking about,
01:06:44:13 – 01:06:45:15
it’s important
01:06:45:15 – 01:06:49:06
that if all of the impact organizations
01:06:49:10 – 01:06:51:09
that were serving the community,
01:06:51:09 – 01:06:55:01
educational, corporate America,
01:06:55:01 – 01:06:55:24
different areas,
01:06:55:24 – 01:06:57:10
we know that part
01:06:57:10 – 01:06:59:18
of our screening process,
01:06:59:18 – 01:07:01:21
people want to sponsor our organization.
01:07:01:21 – 01:07:04:07
They want to to look into this.
01:07:04:07 – 01:07:05:21
They want to look in the screening
01:07:05:21 – 01:07:10:00
is disability access included?
01:07:10:00 – 01:07:11:01
Is that supported?
01:07:11:01 – 01:07:12:10
Are your hiring deaf
01:07:12:10 – 01:07:14:13
and hard of hearing employees?
01:07:14:13 – 01:07:15:14
There’s so many things
01:07:15:14 – 01:07:17:15
to look at discuss with them.
01:07:17:15 – 01:07:19:10
It’s very creative
01:07:19:10 – 01:07:21:12
in the approach to HIPAA.
01:07:21:12 – 01:07:23:17
And so it’s a big it’s a very hot topic.
01:07:25:01 – 01:07:27:16
Not only are the
01:07:27:16 – 01:07:30:05
the community and organizations at large,
01:07:30:05 – 01:07:31:14
but also the individual.
01:07:31:14 – 01:07:32:17
Again,
01:07:32:17 – 01:07:33:21
what it looks like to us
01:07:33:21 – 01:07:35:13
is we have to really partner
01:07:35:13 – 01:07:38:19
and really push this,
01:07:38:19 – 01:07:41:19
that all of them
01:07:42:05 – 01:07:43:08
it’s important for all
01:07:43:08 – 01:07:45:20
because they’re impacted
01:07:45:20 – 01:07:47:18
by the development and design of this.
01:07:47:18 – 01:07:48:05
It’s important
01:07:48:05 – 01:07:49:24
for all people to be included,
01:07:49:24 – 01:07:52:01
not just hire them for, you know,
01:07:52:01 – 01:07:53:10
a consulting position,
01:07:53:10 – 01:07:55:22
a temporary feedback on this,
01:07:55:22 – 01:07:58:00
but have them involved in the development
01:07:58:00 – 01:08:00:20
and design of this. It’s imperative.
01:08:00:20 – 01:08:03:20
It’s so important.
01:08:06:17 – 01:08:08:10
Okay.
01:08:08:10 – 01:08:10:00
For the next question,
01:08:10:00 – 01:08:12:22
automatic speech recognition.
01:08:12:22 – 01:08:16:19
ESR In comparison to eBay.
01:08:16:24 – 01:08:20:08
AI how comfortable
01:08:20:08 – 01:08:22:15
and how much can you trust these?
01:08:22:15 – 01:08:25:15
Two And can you compare the two?
01:08:25:22 – 01:08:28:02
Jeff Here I’d like to take that.
01:08:28:02 – 01:08:31:07
You know, it’s like comparing apples to
01:08:31:07 – 01:08:32:24
oranges, both of them are fruit,
01:08:32:24 – 01:08:34:15
but they are a little bit different.
01:08:34:15 – 01:08:37:22
So this is a very specific portion
01:08:37:22 – 01:08:39:07
of technology
01:08:39:07 – 01:08:42:05
and it focuses on translating from one
01:08:42:05 – 01:08:45:14
language to another.
01:08:45:14 – 01:08:46:17
So so
01:08:48:00 – 01:08:48:19
in its
01:08:48:19 – 01:08:51:20
form, so spoken speech into written form
01:08:52:02 – 01:08:54:24
Abe AI focuses on interpreting
01:08:54:24 – 01:08:57:20
and that is an automatic on the spot
01:08:57:20 – 01:09:00:17
that moment find production.
01:09:00:17 – 01:09:03:03
So that means we have information
01:09:03:03 – 01:09:07:07
feeding into the process
01:09:07:14 – 01:09:10:13
and speech recognition is just picking up
01:09:10:13 – 01:09:12:08
on the kid’s speech.
01:09:12:08 – 01:09:13:14
The context is different,
01:09:13:14 – 01:09:15:18
the information processing is different,
01:09:15:18 – 01:09:17:15
and most of the equivalent
01:09:17:15 – 01:09:20:14
for ASL and speech recognition
01:09:20:14 – 01:09:23:16
would be ASL are signed recognition
01:09:24:01 – 01:09:27:14
and that technology is part of A.I.
01:09:27:14 – 01:09:28:14
by A.I..
01:09:28:14 – 01:09:29:19
But Abe A.I.
01:09:29:19 – 01:09:31:03
includes a variety
01:09:31:03 – 01:09:32:05
other technologies
01:09:32:05 – 01:09:35:05
to help those components work together.
01:09:35:10 – 01:09:37:22
It picks up on that subject
01:09:37:22 – 01:09:40:22
processing, body language,
01:09:41:05 – 01:09:44:01
situational context, clues
01:09:44:01 – 01:09:45:23
and
01:09:45:23 – 01:09:48:23
pragmatics
01:09:49:04 – 01:09:50:13
are the concepts
01:09:50:13 – 01:09:53:13
of the transcriptions.
01:09:53:24 – 01:09:55:05
And just to add to that,
01:09:55:05 – 01:09:56:07
this is Tim here.
01:09:56:07 – 01:10:00:15
And in terms of speech recognition,
01:10:01:16 – 01:10:02:05
that is
01:10:02:05 – 01:10:05:05
something where we already see
01:10:05:15 – 01:10:06:19
a much more development
01:10:06:19 – 01:10:08:11
because there have been years
01:10:08:11 – 01:10:09:05
of investment
01:10:09:05 – 01:10:11:14
in the development of that technology
01:10:11:14 – 01:10:12:19
while sign language
01:10:12:19 – 01:10:14:24
recognition is behind.
01:10:14:24 – 01:10:18:07
And so that becomes an issue of equity.
01:10:18:14 – 01:10:20:01
And there’s a concern there
01:10:20:01 – 01:10:21:17
that with speech recognition,
01:10:21:17 – 01:10:23:20
because of the time and the investment
01:10:23:20 – 01:10:24:22
that’s already there,
01:10:24:22 – 01:10:26:09
we’ve spoken language,
01:10:26:09 – 01:10:28:15
caring people continue to benefit from,
01:10:28:15 – 01:10:29:10
while deaf
01:10:29:10 – 01:10:31:18
individuals cannot have the same access
01:10:31:18 – 01:10:33:09
or the same benefit.
01:10:33:09 – 01:10:35:00
We go back to, for example,
01:10:35:00 – 01:10:37:00
the invention of the telephone.
01:10:37:00 – 01:10:38:04
Of course, hearing
01:10:38:04 – 01:10:39:17
people were able to use the phone
01:10:39:17 – 01:10:42:02
for many years and enjoy that technology
01:10:42:02 – 01:10:43:06
until 80 years later.
01:10:43:06 – 01:10:44:21
We finally got the video phone
01:10:44:21 – 01:10:46:16
and these other forms of technology
01:10:46:16 – 01:10:46:24
where deaf
01:10:46:24 – 01:10:48:05
people could benefit
01:10:48:05 – 01:10:50:18
from the same kind of experience.
01:10:50:18 – 01:10:53:06
So we always have to look at
01:10:53:06 – 01:10:54:08
these situations
01:10:54:08 – 01:10:56:24
and make sure that there is funding
01:10:56:24 – 01:11:00:12
and that research is being done to try to
01:11:00:19 – 01:11:04:08
ensure that sign language is caught up
01:11:04:12 – 01:11:07:08
to what the hearing community is able
01:11:07:08 – 01:11:07:22
to enjoy.
01:11:11:08 – 01:11:12:23
As A next question is
01:11:12:23 – 01:11:14:06
regarding people
01:11:14:06 – 01:11:18:01
with intellectual disabilities, autism
01:11:18:17 – 01:11:21:06
learning disabilities, language
01:11:21:06 – 01:11:24:06
fluency
01:11:24:08 – 01:11:25:07
and the like.
01:11:25:07 – 01:11:28:18
How does this relate to Abe?
01:11:29:01 – 01:11:32:14
I will their abilities and disabilities
01:11:32:14 – 01:11:35:14
be included in this process?
01:11:36:23 – 01:11:38:24
I can take that, Yeah.
01:11:38:24 – 01:11:41:09
Thank you. This is a great question.
01:11:41:09 – 01:11:42:13
Similar
01:11:42:13 – 01:11:44:12
to what we were talking about earlier.
01:11:44:12 – 01:11:47:22
When it comes to, for example,
01:11:48:16 – 01:11:49:24
the FCC,
01:11:49:24 – 01:11:52:13
the Federal Communications Commission,
01:11:52:13 – 01:11:55:13
and talking about language itself
01:11:55:23 – 01:11:58:23
and speech recognition,
01:12:00:21 – 01:12:03:05
there’s understanding
01:12:03:05 – 01:12:05:05
of people
01:12:05:05 – 01:12:06:23
that have different backgrounds.
01:12:06:23 – 01:12:09:24
Same thing happens with this API
01:12:09:24 – 01:12:11:02
and the approach.
01:12:11:02 – 01:12:12:20
We want to make sure that we resolve
01:12:12:20 – 01:12:13:22
some of these issues.
01:12:13:22 – 01:12:14:24
We need to be able
01:12:14:24 – 01:12:16:17
develop something for them
01:12:16:17 – 01:12:19:20
to make sure that it’s successful.
01:12:20:02 – 01:12:23:02
And so it’s still to be seen.
01:12:23:10 – 01:12:24:20
We are not able
01:12:24:20 – 01:12:26:04
answer that at this time
01:12:26:04 – 01:12:28:04
unless there’s anyone else here involved.
01:12:28:04 – 01:12:30:02
But and from what I understand,
01:12:30:02 – 01:12:31:01
I think it’s still a hot
01:12:31:01 – 01:12:32:10
topic of discussion
01:12:32:10 – 01:12:33:12
and it’s something
01:12:33:12 – 01:12:34:17
that the community of people
01:12:34:17 – 01:12:36:13
is still talking about.
01:12:36:13 – 01:12:38:18
But yeah, great question
01:12:38:18 – 01:12:41:12
and saying, yes, Jeff had mentioned the
01:12:41:12 – 01:12:44:18
data, They’re good data and
01:12:45:23 – 01:12:47:00
good models.
01:12:47:00 – 01:12:50:08
So with that design process,
01:12:50:14 – 01:12:51:17
we have to prepare
01:12:51:17 – 01:12:53:09
for a variety of deaf members
01:12:53:09 – 01:12:54:15
in the community.
01:12:54:15 – 01:12:55:00
You know,
01:12:55:00 – 01:12:55:16
in the beginning
01:12:55:16 – 01:12:57:00
we’ve got to collect data
01:12:57:00 – 01:12:58:23
from the community at large
01:12:58:23 – 01:13:01:23
and make sure that it is
01:13:02:09 – 01:13:05:00
appropriate in moving forward
01:13:05:00 – 01:13:05:16
that it’s going to be
01:13:05:16 – 01:13:07:10
beneficial to the community.
01:13:07:10 – 01:13:08:18
Now, if we don’t include
01:13:08:18 – 01:13:09:20
those in the beginning,
01:13:09:20 – 01:13:12:10
that could be a problem with our models.
01:13:12:10 – 01:13:15:10
They won’t be prepared for that.
01:13:17:10 – 01:13:18:19
This is Teresa,
01:13:18:19 – 01:13:21:17
and I’d like to add a specific example
01:13:21:17 – 01:13:23:07
in terms of writing English.
01:13:23:07 – 01:13:25:01
So we see that I know
01:13:25:01 – 01:13:26:10
most people are probably familiar
01:13:26:10 – 01:13:29:10
with Chat. JPT
01:13:29:24 – 01:13:32:12
Currently you’re able to ask chat.
01:13:32:12 – 01:13:36:20
JPT and to develop
01:13:37:00 – 01:13:40:16
some draft of, for example,
01:13:40:16 – 01:13:44:24
plain language meaning the concept
01:13:44:24 – 01:13:46:19
of using it for people
01:13:46:19 – 01:13:49:16
with intellectual disabilities. So
01:13:51:05 – 01:13:52:07
that reflects the
01:13:52:07 – 01:13:53:23
importance of the inclusion
01:13:53:23 – 01:13:56:08
of people in design.
01:13:56:08 – 01:13:57:17
And also
01:13:57:17 – 01:14:00:05
when we are having these discussions
01:14:00:05 – 01:14:03:01
about how to design technology,
01:14:03:01 – 01:14:04:13
we have to ask the question
01:14:04:13 – 01:14:06:06
who’s going to be involved?
01:14:06:06 – 01:14:08:00
Because sometimes the people who are
01:14:08:00 – 01:14:10:01
there are not the people
01:14:10:01 – 01:14:11:01
that need to be involved.
01:14:11:01 – 01:14:12:05
So we want to make sure
01:14:12:05 – 01:14:13:15
that we recognizing
01:14:13:15 – 01:14:14:10
and not forgetting
01:14:14:10 – 01:14:15:16
about these individuals
01:14:15:16 – 01:14:18:16
and these various communities.
01:14:19:08 – 01:14:20:21
I mean, one thing
01:14:20:21 – 01:14:22:03
that I think about is,
01:14:22:03 – 01:14:23:23
for example, a CDI.
01:14:23:23 – 01:14:24:15
Often
01:14:24:15 – 01:14:25:23
we have seen the use
01:14:25:23 – 01:14:27:21
of a certified deaf interpreter
01:14:27:21 – 01:14:30:05
who comes into the situation
01:14:30:05 – 01:14:32:05
and we look at how that changes
01:14:32:05 – 01:14:34:01
and improves communication.
01:14:34:01 – 01:14:35:09
The experience for the deaf
01:14:35:09 – 01:14:37:04
individual is improved,
01:14:37:04 – 01:14:39:02
and I think we can see that benefit
01:14:39:02 – 01:14:40:02
and understand how it would
01:14:40:02 – 01:14:42:10
apply as well to I by I.
01:14:47:02 – 01:14:47:12
Okay.
01:14:47:12 – 01:14:51:03
Next question is related to
01:14:53:01 – 01:14:55:03
there are three different webinars
01:14:55:03 – 01:14:58:03
that you guys hosted
01:14:58:15 – 01:15:01:08
and during those webinars,
01:15:01:08 – 01:15:04:08
did you have a group
01:15:04:13 – 01:15:07:05
or did you have people
01:15:07:05 – 01:15:10:00
who were not using sign language
01:15:10:00 – 01:15:11:24
involved in the group
01:15:11:24 – 01:15:13:19
that they do not use sign language
01:15:13:19 – 01:15:18:00
at all to communicate, to discuss by I
01:15:23:22 – 01:15:26:12
and this is Emery here.
01:15:26:12 – 01:15:28:09
Another good question
01:15:28:09 – 01:15:32:00
and yeah, we had asked them for
01:15:32:12 – 01:15:34:04
to specifically identify
01:15:34:04 – 01:15:37:14
their level of sign skill and
01:15:38:01 – 01:15:39:06
I know that
01:15:39:06 – 01:15:40:19
the panel was talking about that.
01:15:40:19 – 01:15:43:23
I can’t remember exactly how they said,
01:15:43:23 – 01:15:44:16
but it was
01:15:44:16 – 01:15:45:08
the question
01:15:45:08 – 01:15:47:08
was about their level of signing.
01:15:47:08 – 01:15:50:14
And so the it was an open discussion
01:15:50:14 – 01:15:51:21
and we told everyone that
01:15:51:21 – 01:15:53:16
it was going to be in sign language.
01:15:53:16 – 01:15:55:09
But the percentage of people
01:15:55:09 – 01:15:57:07
who were not fluent in
01:15:57:07 – 01:15:59:10
sign, I’m not totally sure.
01:15:59:10 – 01:16:01:22
I don’t know if anyone has a different
01:16:01:22 – 01:16:03:08
recall, something different and
01:16:05:09 – 01:16:05:19
I don’t
01:16:05:19 – 01:16:06:19
think that we had
01:16:06:19 – 01:16:08:19
collected enough data about that.
01:16:08:19 – 01:16:11:15
I think it was hard for us to evaluate,
01:16:11:15 – 01:16:12:14
let alone
01:16:12:14 – 01:16:14:08
evaluate our own data
01:16:14:08 – 01:16:15:17
that we had collected.
01:16:15:17 – 01:16:18:17
So
01:16:21:08 – 01:16:24:08
and in terms of the captioning
01:16:24:08 – 01:16:26:17
and English based caption,
01:16:26:17 – 01:16:28:16
we were not focused on that.
01:16:28:16 – 01:16:29:05
We were more
01:16:29:05 – 01:16:30:19
focused on the sign language
01:16:30:19 – 01:16:32:10
interpreting aspect.
01:16:32:10 – 01:16:36:16
So we do already have a lot of discussion
01:16:36:21 – 01:16:38:00
happening
01:16:38:00 – 01:16:40:22
to the text based aspect of this,
01:16:40:22 – 01:16:43:13
but sign language, there are some gaps,
01:16:43:13 – 01:16:46:13
so we want to focus on that specifically.
01:16:51:00 – 01:16:54:00
Okay,
01:16:55:18 – 01:16:56:17
please help one moment
01:16:56:17 – 01:16:59:17
while through these questions.
01:17:04:07 – 01:17:04:16
Okay.
01:17:04:16 – 01:17:08:00
This question is regarding
01:17:08:00 – 01:17:11:00
the development of the curriculum
01:17:11:07 – 01:17:13:19
at the university level regarding
01:17:13:19 – 01:17:16:19
different interpreting processes
01:17:19:10 – 01:17:22:04
and
01:17:22:04 – 01:17:24:03
for VRA,
01:17:24:03 – 01:17:26:24
VRA and the like.
01:17:26:24 – 01:17:29:21
Now are we going to be adding a I
01:17:31:02 – 01:17:34:01
and there any feedback
01:17:34:01 – 01:17:36:00
that you could share regarding
01:17:36:00 – 01:17:39:12
curriculum development on this subject?
01:17:47:08 – 01:17:48:17
Okay, that’s a good question.
01:17:48:17 – 01:17:49:07
Again,
01:17:49:07 – 01:17:53:09
I think maybe for a future discussion,
01:17:53:09 – 01:17:54:24
it might be more beneficial.
01:17:54:24 – 01:17:55:14
We haven’t
01:17:55:14 – 01:17:57:08
necessarily gotten to that point yet.
01:17:57:08 – 01:17:59:00
We’ll have to do some training
01:17:59:00 – 01:17:59:19
with the curriculum
01:17:59:19 – 01:18:01:10
and workshops and all of.
01:18:01:10 – 01:18:02:03
But again,
01:18:02:03 – 01:18:02:21
I would say
01:18:02:21 – 01:18:06:12
that it could come up at a symposium
01:18:06:22 – 01:18:09:17
or maybe another opportunity
01:18:09:17 – 01:18:13:20
for discussion could become available
01:18:13:20 – 01:18:16:20
in the future.
01:18:21:17 – 01:18:23:24
We haven’t discussed this as of yet,
01:18:23:24 – 01:18:27:05
but in considering the
01:18:27:05 – 01:18:30:05
ethical foundations of this.
01:18:30:10 – 01:18:33:04
So for interpreting,
01:18:33:04 – 01:18:36:16
we do have the ethical foundation
01:18:36:23 – 01:18:38:00
for the Deaf community.
01:18:38:00 – 01:18:42:02
We do have our ethical expectations
01:18:42:20 – 01:18:44:12
and norms
01:18:44:12 – 01:18:46:05
as well as other aspects of that.
01:18:46:05 – 01:18:48:01
But when we enter into this
01:18:48:01 – 01:18:48:24
with some knowledge,
01:18:48:24 – 01:18:52:13
now we have new technology
01:18:52:18 – 01:18:54:07
and what kind of new questions
01:18:54:07 – 01:18:56:05
are going to arise from this?
01:18:56:05 – 01:18:57:17
And that’s part of what we’re hoping
01:18:57:17 – 01:18:59:24
to have some discussions regarding
01:18:59:24 – 01:19:03:01
that topic in the next symposium
01:19:03:01 – 01:19:03:17
next month.
01:19:07:14 – 01:19:08:04
wow, I forgot.
01:19:08:04 – 01:19:09:18
It’s March already.
01:19:09:18 – 01:19:12:18
It is.
01:19:14:16 – 01:19:15:00
Okay.
01:19:15:00 – 01:19:16:08
Next question.
01:19:16:08 – 01:19:19:01
One person had a comment
01:19:19:01 – 01:19:20:13
and said, a lot is happening
01:19:20:13 – 01:19:20:23
right now
01:19:20:23 – 01:19:24:11
in Europe as it relates to AI by AI.
01:19:24:18 – 01:19:28:04
Are you familiar with that and also how
01:19:28:04 – 01:19:32:16
that relates to D GDP?
01:19:33:14 – 01:19:36:11
R And so that’s
01:19:36:11 – 01:19:39:17
the general data protection
01:19:41:01 – 01:19:44:06
regulation to that law.
01:19:44:15 – 01:19:47:20
It’s a very strict
01:19:48:00 – 01:19:51:00
law on privacy and protection.
01:19:51:09 – 01:19:52:08
So can you guys
01:19:52:08 – 01:19:55:08
discuss in touch a bit on that topic?
01:19:56:11 – 01:19:57:12
I would like to do that.
01:19:57:12 – 01:19:59:14
Thank you for bringing that up.
01:19:59:14 – 01:20:04:16
In general, the data has been
01:20:04:18 – 01:20:07:10
it’s been one of the best
01:20:07:10 – 01:20:09:17
data privacy laws.
01:20:09:17 – 01:20:13:07
Hats off to them to the EU
01:20:13:07 – 01:20:14:10
for developing that.
01:20:14:10 – 01:20:15:09
It’s been wonderful
01:20:15:09 – 01:20:17:17
so I think one of the biggest concepts
01:20:17:17 – 01:20:19:02
or take away from this
01:20:19:02 – 01:20:22:02
is the topic of the right
01:20:22:05 – 01:20:24:02
to be forgotten.
01:20:24:02 – 01:20:25:02
Meaning
01:20:25:02 – 01:20:27:06
we can remind them and say, Hey,
01:20:27:06 – 01:20:29:17
I want you to remove my information
01:20:29:17 – 01:20:32:17
and they have to honor your request.
01:20:32:18 – 01:20:35:00
And that is one of the biggest takeaways
01:20:35:00 – 01:20:37:23
for this foundational concept.
01:20:37:23 – 01:20:40:13
And so another thing to look at as
01:20:40:13 – 01:20:43:08
well is the minors.
01:20:43:08 – 01:20:46:08
The age of the data collection.
01:20:46:10 – 01:20:48:21
Parents have to approve that or
01:20:50:03 – 01:20:51:15
revoke approval.
01:20:51:15 – 01:20:52:14
And so really,
01:20:52:14 – 01:20:54:17
in some of these aspects, the EU
01:20:54:17 – 01:20:55:20
ahead of us
01:20:55:20 – 01:20:59:12
and, you know, the American legislature
01:20:59:12 – 01:21:01:22
does have some protections,
01:21:01:22 – 01:21:03:15
but it’s not as focused.
01:21:03:15 – 01:21:06:07
It’s more focused. Children under age.
01:21:06:07 – 01:21:09:07
I believe it’s 13 or 14,
01:21:10:03 – 01:21:11:15
but the EU is ahead of us
01:21:11:15 – 01:21:13:07
in this respect.
01:21:13:07 – 01:21:16:01
The legislation there
01:21:16:01 – 01:21:17:09
there’s a lot of harm
01:21:17:09 – 01:21:19:10
that can be done with this data.
01:21:19:10 – 01:21:20:21
And so that’s something that really needs
01:21:20:21 – 01:21:22:09
to be fleshed out with anyone else.
01:21:22:09 – 01:21:25:09
Like to add to that,
01:21:30:02 – 01:21:31:05
I could go on for,
01:21:31:05 – 01:21:32:23
for ages about this,
01:21:32:23 – 01:21:34:14
but it’s the biggest thing
01:21:34:14 – 01:21:35:15
since
01:21:35:15 – 01:21:39:09
the organization had been established.
01:21:39:09 – 01:21:44:01
The Data collection and privacy
01:21:44:01 – 01:21:47:20
and who is responsible for making sure
01:21:47:20 – 01:21:50:20
that the data is retained safely,
01:21:51:02 – 01:21:53:12
that it’s not leaked?
01:21:53:12 – 01:21:56:16
And if that data is leaked,
01:21:56:18 – 01:21:59:17
how are they going to inform individuals
01:21:59:17 – 01:22:02:13
whose data has been breached that?
01:22:02:13 – 01:22:04:01
This has occurred.
01:22:04:01 – 01:22:05:21
So that’s part of the process
01:22:05:21 – 01:22:07:14
that should be included
01:22:07:14 – 01:22:09:07
in the transparency
01:22:09:07 – 01:22:12:00
and, you know, maintaining that contact
01:22:12:00 – 01:22:15:11
with individuals on the subject.
01:22:16:16 – 01:22:19:16
Thank you.
01:22:21:17 – 01:22:22:04
Okay.
01:22:22:04 – 01:22:25:04
We still have more questions.
01:22:26:02 – 01:22:27:17
This question is related
01:22:27:17 – 01:22:30:17
to machine learning AML.
01:22:30:20 – 01:22:32:24
So and with sign language
01:22:32:24 – 01:22:35:24
recognition, ASL are
01:22:36:19 – 01:22:38:15
who will be training
01:22:38:15 – 01:22:41:15
and teaching the language data.
01:22:41:22 – 01:22:43:17
Where is it going to come from?
01:22:43:17 – 01:22:46:05
From interpreters, from people?
01:22:46:05 – 01:22:49:05
Where will that come from?
01:22:51:01 – 01:22:52:14
So that is an excellent question.
01:22:52:14 – 01:22:53:03
Again,
01:22:53:03 – 01:22:54:07
that is something that
01:22:54:07 – 01:22:56:20
we can’t really control.
01:22:56:20 – 01:22:59:15
It’s up to the company
01:22:59:15 – 01:23:01:18
who is developing that
01:23:01:18 – 01:23:03:06
and doing the work.
01:23:03:06 – 01:23:06:06
So I know that in academia
01:23:06:08 – 01:23:07:19
there’s a lot of evaluations
01:23:07:19 – 01:23:08:24
on authors,
01:23:08:24 – 01:23:12:11
like, for example Oscar Kilner.
01:23:12:19 – 01:23:15:06
01:23:15:06 – 01:23:19:14
Lopez was an author in the ASL,
01:23:19:14 – 01:23:23:14
are community and used interpreter data.
01:23:23:23 – 01:23:27:24
For example, there was a very well-known
01:23:29:03 – 01:23:31:00
individual from Germany
01:23:31:00 – 01:23:34:03
and they had
01:23:35:05 – 01:23:37:11
a a system
01:23:37:11 – 01:23:38:14
that would do
01:23:38:14 – 01:23:40:03
weather reports and alerts
01:23:40:03 – 01:23:41:06
and it had an interpreter
01:23:41:06 – 01:23:42:13
down in the corner.
01:23:42:13 – 01:23:45:17
They recorded that for many years
01:23:46:07 – 01:23:49:07
and they used that data to train
01:23:49:07 – 01:23:51:11
the machine.
01:23:51:11 – 01:23:55:09
And so there was a very limited context
01:23:55:09 – 01:23:57:23
only using that one interpreter.
01:23:57:23 – 01:24:01:02
So it’s a great idea in theory
01:24:01:02 – 01:24:04:02
in the future to use,
01:24:04:19 – 01:24:06:04
you know, in other situations,
01:24:06:04 – 01:24:08:01
it may impact it in that way.
01:24:08:01 – 01:24:11:01
It’s very hard
01:24:11:24 – 01:24:14:11
because the data requires
01:24:14:11 – 01:24:16:16
retention and storage
01:24:16:16 – 01:24:19:16
and a lot of video data available
01:24:19:16 – 01:24:20:10
on the web out
01:24:20:10 – 01:24:23:04
there is not the best quality.
01:24:23:04 – 01:24:24:03
For example,
01:24:24:03 – 01:24:27:03
if you look at like an ASL, one class
01:24:27:05 – 01:24:31:15
student signing hey or a song or whatever
01:24:31:21 – 01:24:33:07
you’re looking and you’re saying,
01:24:33:07 – 01:24:34:07
Hey, you’re doing a good job,
01:24:34:07 – 01:24:35:11
you’re learning, you’re doing well.
01:24:35:11 – 01:24:36:15
But that’s not the model
01:24:36:15 – 01:24:37:22
that we want to use in
01:24:37:22 – 01:24:39:02
training the machines
01:24:39:02 – 01:24:40:03
for machine learning.
01:24:41:04 – 01:24:44:04
And the issue that arises, you know,
01:24:44:07 – 01:24:47:02
is a standard for A.I.,
01:24:47:02 – 01:24:47:12
you know,
01:24:47:12 – 01:24:48:09
where do we get this
01:24:48:09 – 01:24:49:08
data collection from?
01:24:49:08 – 01:24:50:18
From various,
01:24:50:18 – 01:24:51:02
you know,
01:24:51:02 – 01:24:53:06
and that is a bidirectional approach
01:24:53:06 – 01:24:53:24
to interpreting.
01:24:53:24 – 01:24:55:16
But there’s a big problem with that
01:24:55:16 – 01:24:56:18
as well.
01:24:56:18 – 01:24:58:11
Consent and privacy
01:24:58:11 – 01:25:02:01
and confidentiality are highly regulated.
01:25:02:01 – 01:25:04:10
And so we can’t use VR if
01:25:04:10 – 01:25:06:02
even though that would be the best place
01:25:06:02 – 01:25:07:13
for data collection.
01:25:07:13 – 01:25:10:19
So we have both signed and speech
01:25:11:06 – 01:25:13:04
and we are working on
01:25:13:04 – 01:25:15:04
how the two interact
01:25:15:04 – 01:25:19:05
and that’s what is really good.
01:25:19:05 – 01:25:20:06
But we’ve got to look
01:25:20:06 – 01:25:21:11
at different sources
01:25:21:11 – 01:25:23:18
and the organizations themselves.
01:25:23:18 – 01:25:26:12
It’s not
01:25:26:12 – 01:25:28:01
we’ve we’ve got to choose
01:25:28:01 – 01:25:28:22
what data we use.
01:25:28:22 – 01:25:30:16
It’s a very important to start
01:25:30:16 – 01:25:34:07
thinking about our legal framework
01:25:35:12 – 01:25:36:23
and, try to
01:25:36:23 – 01:25:39:23
encourage that and remind them to use
01:25:40:00 – 01:25:40:07
you know,
01:25:40:07 – 01:25:43:19
we’ve got to make sure that the data set
01:25:44:07 – 01:25:48:11
will be include from different sources.
01:25:48:17 – 01:25:50:13
The accuracy is there,
01:25:50:13 – 01:25:52:09
the variety is there.
01:25:52:09 – 01:25:54:06
And the group models,
01:25:54:06 – 01:25:57:22
for example, may be there
01:25:57:22 – 01:26:00:07
may be some fluent individuals,
01:26:00:07 – 01:26:01:14
deaf individuals there.
01:26:01:14 – 01:26:02:13
And that’s not the best
01:26:02:13 – 01:26:03:23
planning model to have.
01:26:03:23 – 01:26:05:03
I know that I’m not the best
01:26:05:03 – 01:26:06:07
signing model to have.
01:26:06:07 – 01:26:08:02
I’m not perfectly fluent myself.
01:26:08:02 – 01:26:09:18
It’s my native language
01:26:09:18 – 01:26:12:10
and I want to understand me.
01:26:12:10 – 01:26:15:02
And the same applies to other
01:26:15:02 – 01:26:16:14
signing styles
01:26:16:14 – 01:26:18:23
and signers with different abilities.
01:26:18:23 – 01:26:20:08
So there’s so many groups
01:26:20:08 – 01:26:21:13
that need to be included
01:26:21:13 – 01:26:24:13
and represented in this machine learning.
01:26:30:01 – 01:26:31:11
And to add to that,
01:26:31:11 – 01:26:32:23
it can be very challenging,
01:26:32:23 – 01:26:35:23
but also it can become an opportunity.
01:26:36:00 – 01:26:36:20
So for example,
01:26:36:20 – 01:26:38:01
we have a lot of deaf
01:26:38:01 – 01:26:39:14
individuals all over
01:26:39:14 – 01:26:42:04
their stories, their history.
01:26:42:04 – 01:26:43:06
A lot of times
01:26:43:06 – 01:26:44:20
this is not shared
01:26:44:20 – 01:26:47:00
and so we want to collect and and
01:26:47:00 – 01:26:48:03
save that data.
01:26:49:22 – 01:26:50:21
So I think it becomes a
01:26:50:21 – 01:26:51:15
bit of a project
01:26:51:15 – 01:26:54:15
to see what the impact is.
01:26:54:15 – 01:26:55:23
Understanding stories
01:26:55:23 – 01:26:58:24
and being able to record this history.
01:26:58:24 – 01:27:00:12
And also it’s an opportunity
01:27:00:12 – 01:27:01:16
to take a deeper dive
01:27:01:16 – 01:27:05:02
into these generational situations and,
01:27:05:11 – 01:27:06:10
getting information
01:27:06:10 – 01:27:08:21
from all over from diverse groups.
01:27:08:21 – 01:27:11:03
But I think the challenge can be funding,
01:27:11:03 – 01:27:13:13
and it’s also challenging to find people
01:27:13:13 – 01:27:14:21
who are able to go out
01:27:14:21 – 01:27:16:21
and record this information
01:27:16:21 – 01:27:18:02
and a quality way.
01:27:18:02 – 01:27:20:20
And so that’s a project in and of itself.
01:27:20:20 – 01:27:22:24
But we have this opportunity now
01:27:22:24 – 01:27:25:24
to look at different domains
01:27:26:03 – 01:27:27:24
and how to use sign language
01:27:27:24 – 01:27:30:17
and the medical and legal legal realms.
01:27:30:17 – 01:27:31:24
And we can see how
01:27:31:24 – 01:27:35:06
that becomes bigger project.
01:27:36:04 – 01:27:39:04
But again,
01:27:40:09 – 01:27:43:12
I think we need to go into that more
01:27:43:12 – 01:27:44:18
and we need to see where
01:27:44:18 – 01:27:46:01
we can get that funding from,
01:27:46:01 – 01:27:47:20
because that’s one of the challenges.
01:27:53:00 – 01:27:53:15
Okay.
01:27:53:15 – 01:27:55:10
Next question.
01:27:55:10 – 01:28:00:02
We have a question from hang on.
01:28:00:02 – 01:28:03:02
I left a deaf interpreter
01:28:05:02 – 01:28:06:02
and the question
01:28:06:02 – 01:28:10:11
is using AI captioning or work
01:28:10:11 – 01:28:13:23
or for a meeting or something of the like
01:28:14:15 – 01:28:15:20
are one of the many tools
01:28:15:20 – 01:28:17:00
that have been tested
01:28:17:00 – 01:28:18:11
with other countries
01:28:18:11 – 01:28:21:11
and their language in their access.
01:28:21:11 – 01:28:23:15
But it seems to fail
01:28:23:15 – 01:28:26:07
or it’s not as accurate
01:28:26:07 – 01:28:28:09
at capturing everything.
01:28:28:09 – 01:28:31:09
What approach would you use for A.I.
01:28:31:10 – 01:28:35:04
by making sure it’s accurate and
01:28:36:06 – 01:28:39:06
successful?
01:28:44:15 – 01:28:45:05
It’s all about
01:28:45:05 – 01:28:48:05
the data collection.
01:28:48:06 – 01:28:49:07
This is the memory here.
01:28:49:07 – 01:28:52:03
I think. Again, a great question.
01:28:52:03 – 01:28:53:14
And to expand on that,
01:28:53:14 – 01:28:55:17
I think we see the same situation
01:28:55:17 – 01:28:57:11
with live captioning
01:28:57:11 – 01:29:01:00
that is nuanced voice tone.
01:29:01:04 – 01:29:04:04
Sign language is just
01:29:04:04 – 01:29:05:12
we have to see how we’re going
01:29:05:12 – 01:29:07:18
to approach that when it comes to AI,
01:29:08:17 – 01:29:10:11
a AI is already challenging
01:29:10:11 – 01:29:12:09
for American Sign language
01:29:12:09 – 01:29:14:03
because it’s a conceptual
01:29:14:03 – 01:29:16:06
and a visual language.
01:29:16:06 – 01:29:18:12
So to be able to capture that
01:29:18:12 – 01:29:21:02
and replicate it and I you know,
01:29:21:02 – 01:29:22:02
this is a great question
01:29:22:02 – 01:29:24:05
because how can we approach this
01:29:24:05 – 01:29:25:15
to make this happen?
01:29:25:15 – 01:29:27:07
It’s still a hot topic.
01:29:27:07 – 01:29:28:17
There’s a lot of discussion on this
01:29:28:17 – 01:29:31:10
because people are still on this process
01:29:31:10 – 01:29:32:14
of trying to screen
01:29:32:14 – 01:29:35:00
and figure out how this pertains to A.I..
01:29:35:00 – 01:29:37:04
But yeah,
01:29:37:04 – 01:29:38:14
Tim, here I’d like to add,
01:29:38:14 – 01:29:40:13
if you look at different fields,
01:29:40:13 – 01:29:43:13
for example, linguistic studies,
01:29:43:16 – 01:29:45:01
the sign language,
01:29:45:01 – 01:29:48:23
it starts in about 1960,
01:29:49:07 – 01:29:51:23
68 with Stokie and his team.
01:29:51:23 – 01:29:53:23
And so you look at ASL
01:29:53:23 – 01:29:55:06
and you look at that
01:29:55:06 – 01:29:57:05
in how it’s developed over time,
01:29:57:05 – 01:29:58:17
and it’s really only been studied
01:29:58:17 – 01:30:01:02
in depth for 50 to 60 years.
01:30:01:02 – 01:30:03:12
It’s in its infancy at best.
01:30:03:12 – 01:30:05:15
And so there’s so many fine languages
01:30:05:15 – 01:30:06:11
all around the world
01:30:06:11 – 01:30:07:17
that have not been studied
01:30:07:17 – 01:30:09:05
and have not been documented
01:30:09:05 – 01:30:10:13
to that degree.
01:30:10:13 – 01:30:12:24
So we still need some more research.
01:30:12:24 – 01:30:15:24
And where you know,
01:30:16:08 – 01:30:17:12
it’s very important
01:30:17:12 – 01:30:19:13
for us to really look at that
01:30:19:13 – 01:30:20:18
with science languages,
01:30:20:18 – 01:30:22:10
but also with many spoken languages
01:30:22:10 – 01:30:23:07
as well.
01:30:23:07 – 01:30:24:24
There’s, you know,
01:30:24:24 – 01:30:26:21
thousands of languages in the world
01:30:26:21 – 01:30:29:06
and they don’t have a written form
01:30:29:06 – 01:30:31:07
for every one of them.
01:30:31:07 – 01:30:32:22
And so
01:30:32:22 – 01:30:34:10
there’s a lot of minority languages
01:30:34:10 – 01:30:36:18
as well and dialects.
01:30:36:18 – 01:30:40:04
And so those are at risk for extinction
01:30:40:11 – 01:30:41:18
because they’re not written,
01:30:41:18 – 01:30:43:07
they’re not studied,
01:30:43:07 – 01:30:44:24
they’re not documented.
01:30:44:24 – 01:30:46:08
And in a few years,
01:30:46:08 – 01:30:47:13
the most common languages
01:30:47:13 – 01:30:50:02
that will be used are major languages.
01:30:50:02 – 01:30:51:06
In the minority languages
01:30:51:06 – 01:30:52:22
will have dissipated.
01:30:52:22 – 01:30:56:03
And so that is something that is a risk.
01:30:56:03 – 01:30:59:23
And we have to look at we can’t leave
01:30:59:23 – 01:31:00:21
those behind.
01:31:04:20 – 01:31:05:03
I’d like
01:31:05:03 – 01:31:06:15
to add to that comment
01:31:06:15 – 01:31:08:24
when we talk about data.
01:31:08:24 – 01:31:10:16
I made a comment recently about that
01:31:10:16 – 01:31:11:03
I was wrong.
01:31:11:03 – 01:31:13:12
I should have said that more data
01:31:13:12 – 01:31:15:08
is coming from
01:31:15:08 – 01:31:17:18
underrepresented communities
01:31:17:18 – 01:31:19:14
and that we need to identify
01:31:19:14 – 01:31:20:20
those communities
01:31:20:20 – 01:31:23:03
and make them aware
01:31:23:03 – 01:31:24:19
that we’d like them to collaborate
01:31:24:19 – 01:31:25:16
with us
01:31:25:16 – 01:31:26:12
to make sure
01:31:26:12 – 01:31:29:12
that they are represented in the data.
01:31:38:24 – 01:31:39:09
Okay.
01:31:39:09 – 01:31:40:20
Tim recently made a comment
01:31:40:20 – 01:31:42:14
about research,
01:31:42:14 – 01:31:45:17
and I’d like to piggyback
01:31:45:17 – 01:31:46:19
my question off of that.
01:31:46:19 – 01:31:49:19
Are there any publications, books,
01:31:49:24 – 01:31:51:14
resources, articles
01:31:51:14 – 01:31:53:21
related to the issue of A.I.
01:31:53:21 – 01:31:58:01
and the Deaf community in Conflict
01:31:59:17 – 01:32:02:10
and I’m sorry, together
01:32:02:10 – 01:32:05:04
intersection.
01:32:05:04 – 01:32:08:03
Can you spell that again?
01:32:08:03 – 01:32:09:20
I’m sorry.
01:32:09:20 – 01:32:10:16
Intersection.
01:32:10:16 – 01:32:13:16
Intersection.
01:32:13:18 – 01:32:16:13
Okay, so my focus in the research
01:32:16:13 – 01:32:19:19
is more on the technical side of things.
01:32:19:19 – 01:32:22:15
I’m not necessarily too heavy on the
01:32:23:14 – 01:32:24:23
on the other
01:32:24:23 – 01:32:26:20
area with the deaf community
01:32:26:20 – 01:32:28:19
and looking into the socio side of it.
01:32:28:19 – 01:32:29:18
But I do think it’s
01:32:29:18 – 01:32:32:21
very interesting research and
01:32:34:02 – 01:32:36:01
I think it’s a good way to do
01:32:36:01 – 01:32:39:08
some formal search of the literature.
01:32:39:14 – 01:32:40:21
So I would suggest
01:32:40:21 – 01:32:42:07
that if you’re interested in this,
01:32:42:07 – 01:32:43:17
that you look at things
01:32:43:17 – 01:32:45:02
like, for example,
01:32:45:02 – 01:32:48:20
Richard Dana or Danny Bragg
01:32:49:04 – 01:32:51:01
and they both have worked
01:32:51:01 – 01:32:54:02
focus on the ethical aspect
01:32:54:22 – 01:32:59:24
of these and the surrounding topics.
01:32:59:24 – 01:33:01:04
And so it’s a lot,
01:33:01:04 – 01:33:02:08
but I could probably
01:33:02:08 – 01:33:06:13
share a bibliography for you all
01:33:06:13 – 01:33:07:16
so that you could take a look
01:33:07:16 – 01:33:08:15
at those authors
01:33:08:15 – 01:33:11:15
and look more into their work.
01:33:13:21 – 01:33:15:14
I wish we could,
01:33:15:14 – 01:33:15:24
you know,
01:33:15:24 – 01:33:17:09
with the advisory group,
01:33:17:09 – 01:33:20:03
we touch on so many things
01:33:20:03 – 01:33:21:20
and I wish we could keep discussing
01:33:21:20 – 01:33:25:01
how the Germans are.
01:33:25:01 – 01:33:26:23
The shared study and research
01:33:26:23 – 01:33:28:04
really impacts everything.
01:33:30:00 – 01:33:30:19
Good idea.
01:33:30:19 – 01:33:33:19
Good idea.
01:33:40:01 – 01:33:40:16
Okay.
01:33:40:16 – 01:33:43:05
Any more questions?
01:33:43:05 – 01:33:45:07
Okay, I do have
01:33:45:07 – 01:33:47:09
I have more questions here,
01:33:47:09 – 01:33:50:09
so I will
01:33:50:23 – 01:33:53:08
copy exactly from what
01:33:53:08 – 01:33:54:11
I’m seeing here on the question,
01:33:54:11 – 01:33:55:15
this might be a good question
01:33:55:15 – 01:33:57:07
for Jeff to answer.
01:33:57:07 – 01:34:00:08
So, Jeff, we’re wondering
01:34:00:08 – 01:34:02:15
if the deaf community
01:34:02:15 – 01:34:05:15
is open to this system.
01:34:06:05 – 01:34:09:05
So for each individual user,
01:34:11:19 – 01:34:14:06
will they all be trained
01:34:14:06 – 01:34:17:19
and taught in their language of ASL?
01:34:17:19 – 01:34:18:14
For example,
01:34:18:14 – 01:34:20:07
if there’s a website,
01:34:20:07 – 01:34:23:07
well, that have American Sign language
01:34:23:09 – 01:34:26:09
and will it be copied and saved,
01:34:26:18 – 01:34:30:07
and will there be anything like
01:34:30:07 – 01:34:31:11
if this is saved,
01:34:31:11 – 01:34:32:16
will personal information
01:34:32:16 – 01:34:34:14
be saved in a server?
01:34:34:14 – 01:34:37:04
How will they ensure confidentiality?
01:34:37:04 – 01:34:37:16
While deaf
01:34:37:16 – 01:34:41:12
individuals have personal rights
01:34:41:12 – 01:34:43:06
to say yes, I’m
01:34:43:06 – 01:34:44:19
okay with releasing my information
01:34:44:19 – 01:34:46:14
at this company, How will that work?
01:34:48:14 – 01:34:50:10
Yeah, that’s a great question.
01:34:50:10 – 01:34:53:04
I think the formal term for
01:34:53:04 – 01:34:56:04
that is called fine tuning.
01:34:57:13 – 01:35:00:08
So with fine tuning you can go in
01:35:00:08 – 01:35:03:13
and make sure that the model is a fit
01:35:03:13 – 01:35:05:12
for your personal style,
01:35:05:12 – 01:35:08:11
your personal terms of choice.
01:35:08:11 – 01:35:11:15
So we can do that currently with English.
01:35:11:15 – 01:35:13:07
When it comes to Elm,
01:35:13:07 – 01:35:15:08
large language models like for example,
01:35:15:08 – 01:35:18:19
Chat, GPT, you can go in and personalize
01:35:18:19 – 01:35:20:01
and bind tune
01:35:20:01 – 01:35:21:23
the data that comes out of it.
01:35:21:23 – 01:35:23:15
And so I would imagine
01:35:23:15 – 01:35:24:09
that the same thing
01:35:24:09 – 01:35:25:09
will eventually happen
01:35:25:09 – 01:35:28:18
when it comes to AI by AI and also
01:35:28:18 – 01:35:31:18
in terms of sharing with others.
01:35:31:23 – 01:35:34:19
I don’t say I don’t see any reason why
01:35:34:19 – 01:35:36:01
it would take a long time
01:35:36:01 – 01:35:36:14
to be able
01:35:36:14 – 01:35:38:07
to for people
01:35:38:07 – 01:35:40:01
to give their informed consent.
01:35:40:01 – 01:35:41:15
So it is your data
01:35:41:15 – 01:35:42:20
and you’ll be able to do
01:35:42:20 – 01:35:44:22
whatever you want to do with it.
01:35:44:22 – 01:35:46:23
That would be the idea.
01:35:46:23 – 01:35:48:02
I don’t know if anyone has
01:35:48:02 – 01:35:49:08
anything else to add to that
01:35:51:14 – 01:35:52:16
Emery here, I
01:35:52:16 – 01:35:54:05
would like to add a comment.
01:35:54:05 – 01:35:55:19
I think that
01:35:55:19 – 01:35:58:19
the community feels empowered
01:35:59:14 – 01:36:02:11
to have these options available.
01:36:02:11 – 01:36:05:00
So I think whether or not
01:36:05:00 – 01:36:07:02
everyone agrees or not,
01:36:07:02 – 01:36:08:18
the idea of having options
01:36:08:18 – 01:36:11:10
available to personalize their data
01:36:11:10 – 01:36:12:23
I think would be optimal
01:36:12:23 – 01:36:15:23
and well received in general
01:36:17:16 – 01:36:19:01
to say, okay, you know,
01:36:19:01 – 01:36:20:15
I see that this is happening,
01:36:20:15 – 01:36:22:10
but I’m just one person.
01:36:22:10 – 01:36:25:01
But that would be my my guess.
01:36:25:01 – 01:36:26:17
This is Tim.
01:36:26:17 – 01:36:28:22
I remember back in what was it, 2001
01:36:28:22 – 01:36:30:11
when I was in college,
01:36:30:11 – 01:36:33:02
we had speech recognition software
01:36:33:02 – 01:36:35:05
and it was called Dragon
01:36:35:05 – 01:36:37:04
Natural Speaking.
01:36:37:04 – 01:36:41:04
And so that required a lot of training,
01:36:41:12 – 01:36:44:24
a lot of feeding into the technology to
01:36:44:24 – 01:36:46:00
develop it.
01:36:46:00 – 01:36:47:22
But there was some control
01:36:47:22 – 01:36:49:18
of sound and clarity.
01:36:49:18 – 01:36:51:24
But I think today the technology
01:36:51:24 – 01:36:54:01
seems to have gotten even better,
01:36:54:01 – 01:36:57:06
and so there’s less of a curve there.
01:36:57:06 – 01:36:59:09
But I think it would be possible
01:36:59:09 – 01:37:01:16
for bias to still be involved.
01:37:01:16 – 01:37:03:09
And so, of course, we see
01:37:03:09 – 01:37:04:16
that it’s quite standard English,
01:37:04:16 – 01:37:06:05
especially in particular tool
01:37:06:05 – 01:37:07:03
I was just discussing.
01:37:07:03 – 01:37:10:08
So when it comes to dialect and accents
01:37:10:08 – 01:37:11:07
and that kind of thing,
01:37:11:07 – 01:37:12:13
for my understanding,
01:37:12:13 – 01:37:14:17
the bigger concern is coming from
01:37:14:17 – 01:37:15:20
the greater community
01:37:15:20 – 01:37:16:14
based on what we see
01:37:16:14 – 01:37:18:03
on speech recognition.
01:37:18:03 – 01:37:20:02
So we can only imagine
01:37:20:02 – 01:37:20:21
what it might look like
01:37:20:21 – 01:37:21:19
with sign language.
01:37:21:19 – 01:37:23:08
So with over 20 years
01:37:23:08 – 01:37:25:11
invested in these types of tools
01:37:25:11 – 01:37:27:10
and and seeing how that goes,
01:37:27:10 – 01:37:28:04
I think that
01:37:28:04 – 01:37:29:23
we need to take into consideration
01:37:29:23 – 01:37:32:04
that we would need better technology
01:37:32:04 – 01:37:35:04
and we would need, you know, that,
01:37:35:14 – 01:37:37:13
for example, leapfrog technology
01:37:37:13 – 01:37:40:13
where we would come in and
01:37:40:20 – 01:37:41:24
be able to say,
01:37:41:24 – 01:37:43:04
we’ve already seen these things
01:37:43:04 – 01:37:45:05
develop, we’ve seen how this has gone.
01:37:45:05 – 01:37:46:22
So maybe we wouldn’t
01:37:46:22 – 01:37:48:23
need as much time to catch up.
01:37:48:23 – 01:37:49:24
I would hope,
01:37:49:24 – 01:37:50:22
like I mentioned, that
01:37:50:22 – 01:37:53:22
that dragon technology’s like 20 years in
01:37:53:22 – 01:37:57:22
so I think that I’m not necessarily
01:37:57:22 – 01:38:00:12
into the technical aspect as much as,
01:38:00:12 – 01:38:01:02
but that’s just
01:38:01:02 – 01:38:02:17
based on my experience time.
01:38:02:17 – 01:38:04:06
I predict it could go.
01:38:07:12 – 01:38:08:24
Holly says the person that asked
01:38:08:24 – 01:38:10:08
that question added a comment
01:38:10:08 – 01:38:12:11
and said, Yes,
01:38:12:11 – 01:38:14:09
I know exactly what you’re talking about.
01:38:14:09 – 01:38:15:09
The dragon, naturally
01:38:15:09 – 01:38:16:10
speaking technology.
01:38:16:10 – 01:38:18:20
I’m familiar with that.
01:38:18:20 – 01:38:20:14
Okay.
01:38:20:14 – 01:38:23:14
Another question says,
01:38:23:23 – 01:38:26:23
I am a deaf leader in my community.
01:38:28:13 – 01:38:32:19
How can we build more of a
01:38:33:04 – 01:38:34:17
bond,
01:38:34:17 – 01:38:37:18
a stronger bond with the interpreters?
01:38:38:03 – 01:38:40:14
Because many of them are afraid
01:38:40:14 – 01:38:43:14
to reach out to the deaf?
01:38:45:01 – 01:38:47:08
And said General question.
01:38:47:08 – 01:38:49:10
Emery says, Can you repeat the question,
01:38:49:10 – 01:38:52:10
please?
01:38:53:08 – 01:38:54:13
It’s a general question.
01:38:54:13 – 01:38:55:16
Holly says,
01:38:55:16 – 01:38:58:16
I am a deaf leader in my community
01:38:59:10 – 01:39:02:06
and I want to know how we as the deaf
01:39:02:06 – 01:39:05:07
community, can build stronger bonds
01:39:06:19 – 01:39:09:18
with interpreters.
01:39:09:18 – 01:39:12:17
Many interpreters afraid,
01:39:12:17 – 01:39:15:17
and they don’t reach out to us and from
01:39:15:17 – 01:39:16:11
the deaf community,
01:39:19:01 – 01:39:21:14
Tim says.
01:39:21:14 – 01:39:24:08
I think that
01:39:24:08 – 01:39:26:18
it’s all comes back to trust
01:39:26:18 – 01:39:28:10
trust issues.
01:39:28:10 – 01:39:29:23
So from my understanding,
01:39:29:23 – 01:39:32:06
many deaf people are afraid
01:39:32:06 – 01:39:34:03
and resistant to technology
01:39:34:03 – 01:39:36:17
because they’re afraid that
01:39:36:17 – 01:39:38:03
this technology
01:39:38:03 – 01:39:39:10
would take away their ability
01:39:39:10 – 01:39:40:20
to have an informed choice
01:39:40:20 – 01:39:42:00
to make decisions.
01:39:42:00 – 01:39:44:03
Same things with same thing with video
01:39:44:03 – 01:39:45:05
remote interpreting.
01:39:45:05 – 01:39:47:04
A lot of deaf people didn’t want that
01:39:47:04 – 01:39:48:00
because,
01:39:48:00 – 01:39:48:05
you know,
01:39:48:05 – 01:39:49:23
they have to fight the system
01:39:49:23 – 01:39:51:22
to get an in-person interpreter.
01:39:51:22 – 01:39:53:11
So based on that experience
01:39:53:11 – 01:39:54:16
and, those challenges,
01:39:54:16 – 01:39:57:16
it’s caused a lot of issues with trust.
01:39:57:19 – 01:40:00:01
And
01:40:00:01 – 01:40:00:16
I know
01:40:00:16 – 01:40:03:03
especially when it comes to health care,
01:40:03:03 – 01:40:04:12
it’s already challenging
01:40:04:12 – 01:40:05:14
to make appointments
01:40:05:14 – 01:40:06:07
to come
01:40:06:07 – 01:40:09:07
in, to have access and all of that.
01:40:09:10 – 01:40:12:20
So it can be very exhausting and cause
01:40:12:20 – 01:40:13:15
deaf individuals
01:40:13:15 – 01:40:14:04
to feel like
01:40:14:04 – 01:40:15:08
they don’t want to go to the doctor
01:40:15:08 – 01:40:16:09
because they don’t want to deal
01:40:16:09 – 01:40:17:14
with all of that.
01:40:17:14 – 01:40:19:17
Now, when it comes to interpreters,
01:40:19:17 – 01:40:23:10
of course, there are some things
01:40:23:10 – 01:40:24:00
to consider,
01:40:24:00 – 01:40:25:21
like the code of ethics, code of coverage
01:40:25:21 – 01:40:26:18
reality.
01:40:26:18 – 01:40:28:13
And I think deaf people may feel,
01:40:28:13 – 01:40:29:09
you know what,
01:40:29:09 – 01:40:32:09
I would prefer to have a I, because
01:40:32:23 – 01:40:35:12
in this case, there’s no baggage.
01:40:35:12 – 01:40:36:15
I don’t have to deal with
01:40:36:15 – 01:40:39:15
the human aspect of trust issues.
01:40:39:16 – 01:40:42:11
I know that with AI
01:40:42:11 – 01:40:45:08
and that kind of collaboration and dialog
01:40:45:08 – 01:40:47:12
also, it can lead to the discussion
01:40:47:12 – 01:40:48:18
of what the meaning
01:40:48:18 – 01:40:50:12
of trust is, what confidentiality
01:40:50:12 – 01:40:51:14
should look like,
01:40:51:14 – 01:40:53:24
the code of ethics, how that applies,
01:40:53:24 – 01:40:55:04
and just to make sure
01:40:55:04 – 01:40:58:02
that if our expectations of AI
01:40:58:02 – 01:41:00:02
are high and
01:41:01:14 – 01:41:02:15
we need to know
01:41:02:15 – 01:41:03:09
those accurate
01:41:03:09 – 01:41:04:20
or what could we expect something
01:41:04:20 – 01:41:06:08
that we’ve had with the human experience.
01:41:06:08 – 01:41:09:08
So that’s another discussion to have.
01:41:10:15 – 01:41:12:03
Theresa I don’t know if you want to add
01:41:12:03 – 01:41:13:01
to that.
01:41:13:01 – 01:41:15:02
Theresa says Yes, I’m just thinking.
01:41:15:02 – 01:41:18:23
I think that maybe the first step
01:41:19:02 – 01:41:21:18
would be to start these discussions
01:41:21:18 – 01:41:24:15
and to have some time in smaller
01:41:24:15 – 01:41:25:16
communities
01:41:25:16 – 01:41:28:08
where everyone knows each other, right?
01:41:28:08 – 01:41:29:11
We are all familiar
01:41:29:11 – 01:41:30:23
with those types of situations
01:41:30:23 – 01:41:32:07
where the interpreters and all the deaf
01:41:32:07 – 01:41:33:04
people and the deaf people,
01:41:33:04 – 01:41:34:07
not the interpreters,
01:41:34:07 – 01:41:37:22
but maybe we would start with some kind
01:41:37:22 – 01:41:39:12
of, let’s say, for example,
01:41:39:12 – 01:41:43:07
have your local deaf
01:41:43:07 – 01:41:47:17
community groups and local or I.D.
01:41:47:19 – 01:41:49:10
or interpreting organizations
01:41:49:10 – 01:41:50:20
come together.
01:41:50:20 – 01:41:51:13
So, for example,
01:41:51:13 – 01:41:52:13
maybe they come together
01:41:52:13 – 01:41:53:18
and watch this video
01:41:53:18 – 01:41:55:17
and then they host a discussion
01:41:55:17 – 01:41:57:02
and ask questions.
01:41:57:02 – 01:41:59:22
But I’m just thinking, how can we start
01:41:59:22 – 01:42:01:17
to develop this discussion
01:42:01:17 – 01:42:03:02
and this dialog?
01:42:03:02 – 01:42:05:05
Because without the dialog,
01:42:05:05 – 01:42:07:08
there’s so many misunderstandings.
01:42:07:08 – 01:42:08:19
And I think that
01:42:08:19 – 01:42:10:19
with the dialog, misunderstandings
01:42:10:19 – 01:42:12:08
will still happen as well,
01:42:12:08 – 01:42:14:06
but is definitely an opportunity
01:42:14:06 – 01:42:14:23
to have
01:42:14:23 – 01:42:16:00
more understanding
01:42:16:00 – 01:42:17:03
and more of an opportunity
01:42:17:03 – 01:42:18:15
to listen to each other
01:42:18:15 – 01:42:20:20
and figure out how we can discuss
01:42:20:20 – 01:42:23:09
this together is huge.
01:42:23:09 – 01:42:24:13
It’s coming
01:42:24:13 – 01:42:26:08
and we need to be sure
01:42:26:08 – 01:42:28:18
that we know how to respond to this.
01:42:28:18 – 01:42:30:08
We need to start with
01:42:30:08 – 01:42:32:24
the grassroots community
01:42:32:24 – 01:42:34:08
and go from there.
01:42:34:08 – 01:42:35:22
So the grassroots
01:42:35:22 – 01:42:36:16
community is the heart
01:42:36:16 – 01:42:37:16
of our deaf community.
01:42:43:04 – 01:42:44:04
Thank you, Anne Marie.
01:42:44:04 – 01:42:44:24
And Jeff, would you
01:42:44:24 – 01:42:46:13
do you have anything you want to add?
01:42:46:13 – 01:42:49:13
Just saying I agree wholeheartedly.
01:42:50:04 – 01:42:51:20
Emery Here,
01:42:51:20 – 01:42:55:18
this one topic itself is just it.
01:42:56:03 – 01:42:57:09
It’s huge.
01:42:57:09 – 01:42:59:09
There’s no way to describe it.
01:42:59:09 – 01:43:00:05
Otherwise,
01:43:00:05 – 01:43:01:21
there are so many things
01:43:01:21 – 01:43:04:17
to look at the process itself,
01:43:04:17 – 01:43:08:02
the trust in the process, the transport
01:43:08:03 – 01:43:11:03
tenancy, the data collection, all of it
01:43:11:06 – 01:43:12:17
together.
01:43:12:17 – 01:43:13:17
But you know,
01:43:13:17 – 01:43:16:06
and how everything moves in tandem.
01:43:16:06 – 01:43:17:06
It’s an opportunity
01:43:17:06 – 01:43:20:06
to always create a safe space
01:43:20:23 – 01:43:23:06
for the community to come together
01:43:23:06 – 01:43:24:17
and to discuss.
01:43:24:17 – 01:43:25:19
And it’s important
01:43:25:19 – 01:43:26:08
for us
01:43:26:08 – 01:43:29:08
to look at that process as a whole.
01:43:31:13 – 01:43:33:04
Very good discussion.
01:43:33:04 – 01:43:35:11
Very good discussion,
01:43:35:11 – 01:43:35:24
Ali, saying,
01:43:35:24 – 01:43:38:24
okay, next question is about the research
01:43:38:24 – 01:43:42:14
process from last October.
01:43:42:14 – 01:43:45:14
You had three different webinar sessions,
01:43:45:18 – 01:43:48:15
the deaf participants in those sessions.
01:43:48:15 – 01:43:49:24
What were their backgrounds,
01:43:49:24 – 01:43:52:19
where were they from geographically?
01:43:52:19 – 01:43:54:07
And
01:43:55:22 – 01:43:58:22
demographically?
01:43:59:03 – 01:44:00:20
Just saying, go ahead.
01:44:00:20 – 01:44:01:07
Emery
01:44:01:07 – 01:44:03:21
So the backgrounds were very diverse.
01:44:03:21 – 01:44:05:18
We had some that were deaf interpreters,
01:44:05:18 – 01:44:07:13
deaf consumers,
01:44:07:13 – 01:44:11:05
deaf individuals who are professionals
01:44:11:10 – 01:44:14:08
working in the field of education,
01:44:14:08 – 01:44:16:07
working in the field of interpreting
01:44:16:07 – 01:44:17:21
very different varieties
01:44:17:21 – 01:44:20:21
of backgrounds and
01:44:21:12 – 01:44:23:04
areas as well.
01:44:23:04 – 01:44:24:22
Jeff Did you want to add some more?
01:44:24:22 – 01:44:25:20
Jeff Yes,
01:44:25:20 – 01:44:26:22
I think the next step
01:44:26:22 – 01:44:28:22
is to decide how we expand
01:44:28:22 – 01:44:30:11
and how we grow our audience
01:44:30:11 – 01:44:32:05
from those webinars
01:44:32:05 – 01:44:34:00
and to really include
01:44:34:00 – 01:44:35:21
even more of the community at large,
01:44:35:21 – 01:44:38:21
to have all those perspectives as well
01:44:40:06 – 01:44:41:09
in the saying yes,
01:44:41:09 – 01:44:42:23
the webinars were really
01:44:42:23 – 01:44:47:15
our first step into this
01:44:47:15 – 01:44:49:19
realm of testing out this,
01:44:49:19 – 01:44:50:21
looking at different things,
01:44:50:21 – 01:44:52:02
and this symposium
01:44:52:02 – 01:44:53:04
will just continue
01:44:53:04 – 01:44:55:21
to be a springboard into the future
01:44:55:21 – 01:44:57:17
and as long as we have a
01:44:57:17 – 01:44:59:13
I will be having these discussions
01:45:00:24 – 01:45:01:16
in raising.
01:45:01:16 – 01:45:02:11
I’d like to add
01:45:02:11 – 01:45:06:05
also that all of the individuals,
01:45:06:13 – 01:45:08:03
the participants here today,
01:45:08:03 – 01:45:09:16
you guys are critical
01:45:09:16 – 01:45:11:16
for this process as well.
01:45:11:16 – 01:45:13:01
We are not finished now
01:45:13:01 – 01:45:14:15
that this report is published.
01:45:14:15 – 01:45:16:05
This is just the beginning.
01:45:16:05 – 01:45:18:12
US and even all of your questions
01:45:18:12 – 01:45:20:04
have really spurred our thoughts
01:45:20:04 – 01:45:22:04
into how we move this forward
01:45:22:04 – 01:45:23:04
and what the next steps
01:45:23:04 – 01:45:24:06
are in this process.
01:45:24:06 – 01:45:27:06
We’ve just scratched the surface.
01:45:34:04 – 01:45:35:21
Okay, I’ll help you think.
01:45:35:21 – 01:45:36:23
We have navigated
01:45:36:23 – 01:45:38:06
through all of the questions
01:45:38:06 – 01:45:39:22
that we have for the day.
01:45:39:22 – 01:45:45:02
We do have one more, but it is
01:45:45:07 – 01:45:46:17
with this
01:45:46:17 – 01:45:49:17
webinar, complete this report, complete
01:45:49:21 – 01:45:52:01
the sharing of the recording
01:45:52:01 – 01:45:53:10
When will that be done?
01:45:53:10 – 01:45:57:08
The Deaf Safe
01:45:57:16 – 01:46:01:10
Advisory website, the report,
01:46:01:16 – 01:46:03:05
if that will be shared,
01:46:03:05 – 01:46:06:09
and how to register
01:46:06:09 – 01:46:09:09
for the Brown University Symposium.
01:46:09:16 – 01:46:10:14
Several people have asked
01:46:10:14 – 01:46:12:06
for that information as well.
01:46:14:18 – 01:46:17:11
Tim here, so I will answer that.
01:46:17:11 – 01:46:21:10
The report we have both the safe
01:46:21:18 – 01:46:25:19
I their report as well as ours,
01:46:25:19 – 01:46:27:00
and we’ve been working
01:46:27:00 – 01:46:29:08
with Katha research.
01:46:29:08 – 01:46:31:10
They have spent so much time
01:46:31:10 – 01:46:32:21
going through this project.
01:46:32:21 – 01:46:36:04
The surveys, volunteers, so much work
01:46:36:04 – 01:46:37:13
has gone into that report
01:46:37:13 – 01:46:38:04
and that will
01:46:38:04 – 01:46:40:00
and will be published as well.
01:46:40:00 – 01:46:41:15
We had a tech
01:46:41:15 – 01:46:43:12
a presentation scheduled yesterday,
01:46:43:12 – 01:46:45:11
but there was technological problems
01:46:45:11 – 01:46:47:00
and so we’re going to be rescheduling
01:46:47:00 – 01:46:47:24
that for Wednesday morning,
01:46:47:24 – 01:46:50:24
I believe, at 11 Eastern time.
01:46:51:00 – 01:46:54:02
And so with that being said,
01:46:54:02 – 01:46:55:13
I really encourage you all
01:46:55:13 – 01:46:57:12
to watch that presentation.
01:46:57:12 – 01:46:59:08
It’s about languages in general,
01:46:59:08 – 01:47:00:14
not just sign languages,
01:47:00:14 – 01:47:01:23
it’s languages in general.
01:47:01:23 – 01:47:03:17
And so that will be made
01:47:03:17 – 01:47:05:10
available to the public as well.
01:47:05:10 – 01:47:06:06
This one,
01:47:06:06 – 01:47:09:17
I believe we have some editing to do
01:47:09:23 – 01:47:12:10
and we will make it available here soon.
01:47:12:10 – 01:47:13:10
The presentation
01:47:13:10 – 01:47:14:01
next Wednesday
01:47:14:01 – 01:47:16:07
will also be available online.
01:47:16:07 – 01:47:18:22
Our website does have links
01:47:18:22 – 01:47:21:22
to Safe A.I.
01:47:22:01 – 01:47:23:11
Advisory Group,
01:47:23:11 – 01:47:25:04
and then we have our own Death
01:47:25:04 – 01:47:27:04
Advisory Group page
01:47:27:04 – 01:47:29:02
and the links to those.
01:47:29:02 – 01:47:30:08
I will thin them out
01:47:30:08 – 01:47:32:07
and make it available there
01:47:32:07 – 01:47:34:16
also the symposium
01:47:34:16 – 01:47:37:16
we are currently working on that platform
01:47:37:20 – 01:47:38:22
and the save
01:47:38:22 – 01:47:40:12
the date is just announced today.
01:47:40:12 – 01:47:42:16
This is our first announcement for that
01:47:42:16 – 01:47:44:19
I will send out a more formal
01:47:44:19 – 01:47:45:14
save the date
01:47:45:14 – 01:47:47:22
with information on registration
01:47:47:22 – 01:47:49:02
and the like.
01:47:49:02 – 01:47:50:05
It will be sent out.
01:47:50:05 – 01:47:52:08
That is a currently in process.
01:47:52:08 – 01:47:53:19
So Look forward to that.
01:47:55:06 – 01:47:55:21
I do.
01:47:55:21 – 01:47:56:20
Any of the other
01:47:56:20 – 01:47:58:06
advisory council members
01:47:58:06 – 01:47:58:21
have something
01:47:58:21 – 01:48:01:10
that they would like to add.
01:48:01:10 – 01:48:02:05
Theresa Thing
01:48:02:05 – 01:48:03:23
I just want to thank all of you
01:48:03:23 – 01:48:06:12
and Mary saying, yes, I agree. Thank you.
01:48:06:12 – 01:48:08:09
Thank you for your interest
01:48:08:09 – 01:48:09:04
in this topic.
01:48:09:04 – 01:48:10:04
Thank you for coming
01:48:10:04 – 01:48:12:03
and listening to our presentation.
01:48:12:03 – 01:48:13:15
We Appreciate it.
01:48:13:15 – 01:48:15:09
We we think that it’s wonderful
01:48:15:09 – 01:48:16:19
that all of you were involved
01:48:16:19 – 01:48:18:22
and we really appreciate everyone
01:48:18:22 – 01:48:20:12
that was involved in the study
01:48:20:12 – 01:48:21:04
just said yes.
01:48:21:04 – 01:48:24:04
Thank you so much.
01:48:24:11 – 01:48:25:17
Thank you, Holly, as well
01:48:25:17 – 01:48:26:21
for your time today.
01:48:26:21 – 01:48:29:06
We appreciate you joining us
01:48:29:06 – 01:48:32:06
in this presentation.
01:48:32:07 – 01:48:35:07
Thank you very much,
01:48:37:06 – 01:48:37:12
Holly.
01:48:37:12 – 01:48:37:19
Same.
01:48:37:19 – 01:48:39:09
I want to make sure that we have a clear
01:48:39:09 – 01:48:40:13
answer about
01:48:40:13 – 01:48:40:22
01:48:40:22 – 01:48:43:10
if this recording of the webinar
01:48:43:10 – 01:48:44:22
will be broadcast
01:48:44:22 – 01:48:46:09
and then
01:48:46:09 – 01:48:47:17
we will be sending up
01:48:47:17 – 01:48:49:06
a follow up email as well.
01:48:49:06 – 01:48:49:19
For everyone
01:48:49:19 – 01:48:50:16
who registered
01:48:50:16 – 01:48:53:16
with the website information
01:48:53:16 – 01:48:56:24
and Brown University Symposium
01:48:58:07 – 01:48:59:23
information
01:48:59:23 – 01:49:03:02
and what else was there.
01:49:03:02 – 01:49:03:24
I think that’s it.
01:49:03:24 – 01:49:06:24
So we will correct?
01:49:07:15 – 01:49:08:21
Yes, simply yes,
01:49:08:21 – 01:49:09:24
we will all just say yes,
01:49:09:24 – 01:49:12:24
it will be all available.
01:49:15:22 – 01:49:18:15
The webinar also that we have,
01:49:18:15 – 01:49:19:23
we have the video
01:49:19:23 – 01:49:21:03
recordings of the webinars,
01:49:21:03 – 01:49:22:13
if you’d like to see those as well.
01:49:22:13 – 01:49:25:13
We those available.
01:49:29:23 – 01:49:31:24
All right,
01:49:31:24 – 01:49:33:18
Tim thing, I believe this concludes
01:49:33:18 – 01:49:36:18
our meeting.
01:49:38:24 – 01:49:40:10
Just saying Thank you, everyone.
01:49:40:10 – 01:49:42:06
We appreciate your time.
01:49:42:06 – 01:49:43:18
Thank you so much. Bye bye.
In the world of AI, what might be called “small language models” have been growing in popularity recently because they can be run on a
In a March 21, 2024 paper, Fan Zhou and Vincent Vandeghinste from KU Leuven demonstrated that language models can predict the most suitable translation techniques for translation and post-editing tasks.
Like most patent attorneys, I get multiple emails each month for artificial intelligence tools purporting to help patent attorneys draft patent applications. I have done
AI auditing is a rapidly growing field of research and practice. This review article, which doubles as an editorial to Digital Society’s topical collection on
Large language models are well versed in Standard American English and a few other dominant world languages where training data is plentiful, but how do
Sunnyvale residents who don’t speak English have a new way to engage and participate in city meetings. The city is piloting an artificial intelligence-based translation
TAUS is now offering its comprehensive data collection of close to 7.4 billion words for sale at discounts of more than 97% off the original value.
There’s been no shortage of new claims and ideas about what generative AI (GenAI) can, cannot, and should not do. And despite the hype, there
The American Translators Association (ATA) is among a diverse group of members of the independent group Stakeholders Advocating for Fair and Ethical AI in Interpreting
Microsoft plans to invest $1.5 billion in an Abu Dhabi-based artificial intelligence company, a deal that could limit China’s influence in the Gulf region amid
Which A.I. system writes the best computer code or generates the most realistic image? Right now, there’s no easy way to answer those questions. Read
A user could ask ChatGPT to write a computer program or summarize an article, and the AI chatbot would likely be able to generate useful