13:39:22 <RRSAgent> RRSAgent has joined #rqtf
13:39:26 <RRSAgent> logging to https://www.w3.org/2025/02/19-rqtf-irc
13:39:26 <Zakim> RRSAgent, make logs Public
13:39:27 <Zakim> please title this meeting ("meeting: ..."), jasonjgw
13:39:30 <jasonjgw> meeting: RQTF meeting
13:39:37 <jasonjgw> chair: jasonjgw
13:39:39 <jasonjgw> scribe+
13:39:41 <jasonjgw> present+
13:40:01 <jasonjgw> agenda+ Accessibility of machine learning and generative AI.
13:55:23 <Joshue108> Joshue108 has joined #rqtf
14:01:18 <janina> janina has joined #rqtf
14:01:22 <janina> present+
14:04:44 <jasonjgw> zakim, next item
14:04:44 <Zakim> agendum 1 -- Accessibility of machine learning and generative AI. -- taken up [from jasonjgw]
14:06:52 <Joshue108> JW: We need to discuss this
14:07:07 <Joshue108> JW: We need a structure on the issues and happy to hear update
14:07:35 <Joshue108> Scribenick: Joshue108
14:08:04 <Joshue108> https://w3c.github.io/ai-accessibility/
14:09:04 <scott_h> scott_h has joined #rqtf
14:09:11 <Joshue108> JW: I've suggested structure on list
14:09:30 <Joshue108> JW: We are not likely to identify all or even most relevant use cases
14:09:41 <Joshue108> We can treat the cases we know about as examples and raise issues
14:09:41 <scott_h> present+
14:09:51 <Joshue108> This would help with a conceptual framework
14:09:55 <Joshue108> +1 to Jason
14:10:13 <Joshue108> JW: This would help the document age well - without the need for constant updates
14:10:20 <Joshue108> Leads to a more enduring approach.
14:10:23 <janina> q+
14:10:35 <Joshue108> +1 from Janina and Scott
14:10:37 <Joshue108> ack jan
14:10:59 <Joshue108> JS: We can point to things in various domains and aspects
14:11:09 <Joshue108> SH: Good point - its too vast.
14:11:17 <Joshue108> We can pick on examples and case studies
14:11:24 <Joshue108> JS: Can we brainstorm?
14:11:29 <Joshue108> Re: Table of contents
14:11:51 <Joshue108> JS: We have the person with a disability on the web trying to take advantage of AI enabled contents
14:11:51 <stacey> stacey has joined #rqtf
14:11:55 <stacey> present+
14:11:57 <Joshue108> We need considerations related to that.
14:12:06 <Joshue108> We have content creation being assisted by AI
14:12:16 <Joshue108> we need to be aware of that - good for text
14:12:30 <Joshue108> Problematic for images - what degree of confidence can we have?
14:12:38 <Joshue108> Code generation is also to be included
14:12:50 <Joshue108> Then the user interacting with various agents - smart travel.
14:12:56 <Joshue108> What will make this helpful?
14:13:36 <Joshue108> JS: How can AI bring accomodations?
14:13:46 <Joshue108> Then there is the user interaction with their own systems.
14:14:17 <Joshue108> I wouldn't mind if it was monitoring what I was doing and helping me be smarter on things that I could do.
14:14:24 <Joshue108> Or notify me from time to time?
14:14:42 <Joshue108> Can vary from user agent. Rich vein is thinking of this.
14:14:52 <Joshue108> Assistive Technologies should be looking at this.
14:15:15 <Joshue108> SH: Saw some smart home/car videos are interesting, immersive AI
14:15:31 <Joshue108> SH: Very granular - and the a11y conversation is not there yet
14:15:48 <Joshue108> JS: The standard agent for folks without a11y needs will not work
14:16:09 <Joshue108> It needs to be considered or it will be an inaccessible world all over again.
14:17:11 <Joshue108> https://w3c.github.io/ai-accessibility/
14:17:45 <Raja_Kushalnagar> Raja_Kushalnagar has joined #rqtf
14:18:20 <Joshue108> JS: We are going to build the TOC shortly and get an a FPWD out shortly
14:18:31 <Joshue108> JS: Would like to see this published soon
14:18:48 <Joshue108> JS: Thanks Janina - these suggestions fall well into my analysis
14:19:18 <Joshue108> We have ML learning in the authoring env, as a provider of app functionality, and then its use to enhance a11y of UI
14:19:37 <janina> q?
14:19:41 <janina> q+
14:19:56 <Joshue108> JOC: Above comments from Jason White btw
14:20:09 <Joshue108> JW: Hope this agreeable
14:20:22 <Joshue108> JW: Suggests working on these contents
14:20:49 <Joshue108> JS: I want to move on with that job - I dont disagree with the categories. Not sure its the top level of the taxonomy
14:21:00 <Joshue108> The user needs to be the centre of everyting
14:21:09 <Joshue108> s/everyting/everything
14:21:31 <Joshue108> JS: All the usual cliches need to be removed
14:22:04 <Joshue108> JS: The hyperbole is exaggerated
14:22:05 <scott_h> LG AI video https://www.youtube.com/watch?v=gYRM00Oe2BM
14:22:15 <Joshue108> JW: Cory doctorow has good article
14:22:34 <Joshue108> JW: There will be good writing on AI A11y - Jason is working also on interesting stuff
14:22:50 <scott_h> LG video shows AI integraiton across multiple devices and user agents but accessibility not considered
14:22:57 <Joshue108> RK: There is a lot of potential with new ways of using AI - advantage and risk needs to be considered
14:22:59 <Joshue108> q+
14:23:02 <Joshue108> The ack jan
14:23:43 <Joshue108> RK: The errors need to be considered, disability tax e.g.
14:24:28 <Joshue108> Disability tax is the increase of work and effort because of a disability in your everyday life
14:24:47 <Joshue108> RK: This needs to be authored in partnership
14:25:04 <Joshue108> As a deaf person, voice detection and captions can be added.
14:25:22 <Joshue108> But there is more work, as it is not in Sign - so there is another layer of translation
14:25:30 <Joshue108> Additional work needs to be thought of
14:25:37 <Joshue108> q?
14:25:41 <Joshue108> ack janina
14:25:46 <Joshue108> +1 to Raja
14:26:23 <Joshue108> SSG: There is also a 'real' tax to have equity - the tools that are needed, medical care etc
14:26:57 <Joshue108> SH: We have noticed this in Oz - things that are 'special'  or disability supported are more expenses
14:27:15 <Joshue108> ack me
14:27:20 <janina> scribe+
14:27:38 <janina> Joshue108: +1 to Raja, agree on tyhese scenarios
14:27:47 <janina> Joshue108: agree also with user first
14:27:56 <janina> Joshue108: This is our opportunity to flag this
14:28:10 <janina> Joshue108: the overlays are built by nurds who don't understand the userneed
14:28:22 <janina> Joshue108: this is important to flag in this doc
14:28:55 <Joshue108> JW: Jason, I'm hearing a concensus as from the user perspective as the top level taxonomyu
14:29:08 <Joshue108> s/taxonomyu/taxonomy
14:29:36 <Joshue108> JW: I think that is reasonable to a large extent but there are issues around the authoring content
14:29:42 <Joshue108> Error handling for example
14:29:44 <Joshue108> q+
14:29:47 <Joshue108> ack me
14:29:58 <Joshue108> q+ to say why cant we do both
14:30:04 <janina> q+
14:30:08 <Joshue108> ack me
14:30:08 <Zakim> Joshue, you wanted to say why cant we do both
14:30:16 <janina> Joshue108: Why can't we do both?
14:30:25 <janina> Joshue108: Agree with jason's points, very valid
14:30:50 <janina> Joshue108: AI now a panacea, the devil is in the details; we're still in the sales stage
14:31:06 <janina> Joshue108: We should ask these larger questions
14:31:07 <janina> q?
14:31:14 <Joshue108> ack jan
14:31:29 <Joshue108> JS: My comments are pragmatic - I want to know the structure
14:31:42 <Joshue108> Am concerned about scope/size
14:33:17 <Joshue108> JOC: I dont see the suggestions that Jason made are antithetical to putting the user first
14:34:12 <Joshue108> JW: I dont think user needs capture it well for this document
14:34:21 <Joshue108> We have other areas to cover
14:34:28 <Joshue108> JS: What is an example?
14:34:56 <Joshue108> JW: Ok, regarding content creation - you are satisfying user needs for abstract users at that point
14:35:08 <Joshue108> Under application development that could be a heading
14:35:35 <Joshue108> JS: Counter argument - that sounds like a philosophical thing - sounds like user needs to me
14:36:50 <Joshue108> SH: This work started on talking about their perspective in AI - there are other things - remedial actions, how to engage on AI to support People with disabilities and to Josh point about how can we trust it and its ability to support users with disabiities
14:37:07 <Joshue108> SH: Those elements need to come in
14:37:17 <Joshue108> JW: I have to agree with Janina on this
14:37:38 <Joshue108> e.g. the authoring process does support people with disabilities -
14:37:41 <Joshue108> JW: How?
14:38:03 <Joshue108> JS: Difference between quality management
14:38:12 <Joshue108> +1 to Jason - quality is critical here
14:38:26 <Joshue108> JW: Yes, this will be error prone tech for a long time
14:39:05 <Joshue108> JS: Just to be clear - in the intro we can say this tech will be error prone, some of this will produce content that wont work
14:39:11 <Joshue108> I don't want to discuss that
14:39:19 <Joshue108> JW: I think this is critical
14:39:21 <Joshue108> q+
14:39:22 <Joshue108> ack me
14:39:50 <Joshue108> q+ to say that the potential scope of this is so big these conversations are important for us
14:40:03 <Joshue108> RK: The question about authoring ..
14:40:18 <Joshue108> Creating videos and content - what are we using the term authorship?
14:40:40 <Joshue108> JS: I think Jason is thinking very broadly, doctoral work, cool app building etc
14:40:44 <Joshue108> JW: Yes
14:41:35 <Joshue108> <Beethoven is being invoked>
14:41:49 <Joshue108> SH: Who we writing for?
14:41:59 <Joshue108> JOC: Great question
14:42:15 <Joshue108> SH: Devs and designers that engage with the user?
14:43:20 <janina> jasonjgw: Anyone trying to integrate AI into their applications; code,, AT, content, etc
14:43:30 <janina> jasonjgw: Highly relevant audiences for this work
14:43:45 <janina> jasonjgw: We want to influence those people in W3C and elsewhere
14:43:59 <janina> scott_h: To support a user need? or other applications?
14:44:25 <janina> scott_h: So errors, about ML to pick up errors in ml itself? Is it its own end?
14:44:43 <janina> jasonjgw: Well, that's useful, but in the end that needs to serve greater a11y
14:45:05 <janina> Raja_Kushalnagar: Currently no way for AI to catch its own errors
14:45:14 <janina> Raja_Kushalnagar: maybe someday, but not yuet
14:45:23 <janina> s/yuet/yet/
14:45:40 <janina> jasonjgw: We have error handling as a major problem
14:45:55 <janina> scott_h: Yes, ai halucinations ...
14:46:09 <janina> Raja_Kushalnagar: Showing to the user and the audience
14:46:48 <janina> Raja_Kushalnagar: Sometimes we'll see things -- in captions
14:46:48 <janina> Raja_Kushalnagar: Both people involved need to see them
14:46:52 <janina> scott_h: If AI generates someone with 6 fingers, a low vision person might not catch
14:47:13 <Joshue108> We also have the potential for AI to produce trashy code that could be sold as being accessible remediation
14:47:19 <stacey> q+
14:47:19 <janina> jasonjgw: when ai is generating part or all of the ui ... not scripted -- all ml generated
14:47:35 <janina> jasonjgw: what capability makes the user's a11y needs supported
14:48:07 <janina> jasonjgw: errors an issue there too, but can it meet my cognitive needs, my at needs
14:48:27 <Joshue108> q-
14:49:12 <janina> Joshue108: also a11y remediation ...
14:53:25 <janina> jasonjgw: Two things we've identified:
14:53:48 <janina> jasonjgw: 1 Issues of accuracy in general -- accuracy of what the ML does
14:54:40 <janina> jasonjgw: 2. what are the capabilities of the ml when it's producing content and/or ui -- does it comprhend my a11y need and accomodate
14:56:53 <jasonjgw> zakim, end meeting
14:56:53 <Zakim> As of this point the attendees have been jasonjgw, janina, scott_h, stacey
14:56:55 <Zakim> RRSAgent, please draft minutes
14:56:57 <RRSAgent> I have made the request to generate https://www.w3.org/2025/02/19-rqtf-minutes.html Zakim
14:57:04 <Zakim> I am happy to have been of service, jasonjgw; please remember to excuse RRSAgent.  Goodbye
14:57:04 <Zakim> Zakim has left #rqtf
14:58:16 <janina> janina has left #rqtf