Meeting minutes
Accessibility of Machine Learning and Generative AI draft.
Jason: Janina circulated a doc with proposed TOC and under each section there's a series of descriptive paragraphs that introduce what the issue is. Developed with aid of LLM. Thanks to Josh, Stacey, and Scott for sending comments to the group. Janina - first question - given comments so far it's broadly supported with detailed suggestions.
Thoughts on your side?
Janina: I was waiting for this meeting and not looking. Will do so now that there are comments to respond to. Bottom line - is this a path forward in a useful direction? And text in each area, worth keeping, reflects what we said and all needs tweaking and too long. But if general concensus is it's step forward then I'll proceed and incirporate
what we had so far in place. Caveat - I'll come back to that. So one doc in Git in our usual structure and start trimming discussion in how we got to this. Leave short paragraph of what we want to say in doc development. Caveat - we have a section on overlays. I don't think it belongs. Scripted responses, not usually AI.
Jason: Can make it clear it's an AT or app on website, and probably enough of what to say
Janina: AI can be applied anywhere in the process and analysis of the doc...helpful in understanding where things happen in the process when someone puts content online and a user with a disability needs to interact with it. But that's done after the fact in a user agent, publishing...series of places where you want a11y concerns addressed. They
can be addressed well or poorly. Ways to address appropriately. Original overlay was a bad outcome and that's (the fact sheet) I think it doesn't belong and can get in our way to reasonably discuss things.
Jason: issue that doesn't touch on ML technologies too much except can be a mechanism, trajectory from creation to usage. I think I'd agree with you
Janina: All dynamic. AI makes it more so
Jason: UI elements that are AI generated. Totally a different scenario. What you proposed TOC I think captures many more issues we've discussed than existed structure and better starting point than what we had. And use as starting point.
… any objections? [no]
Janina: so they are perfecting comments not on the approach?
Jason: yes
… actions to suggest: first is to take into account comments that have been made on mailing list. Detail level. Then second, further refinement and I can comment on any stage along the way. I think my comments would be at a detailed level anyway. Sections/organization. So no comments that would hold up the work. If you're satisfied that you know
what you're doing and we're supporting you I think it takes care of this one
Janina: I can progress and move forward reasonably well
Jason: any final questions or comments before we revisit? Maybe next week. Might be more as the work goes on.
Janina: Next step is to create git file and get us a branch
Jason: and opportunity to comment on that. I can do comments at that stage.
Janina: branch or retire current doc into an earlier draft? Branch still makes sense
Jason: makes sense. Merge would mostly replace what was there
janina: yes
Janina: OK, will think in terms of a branch
[no final comments and questions]
Metaverse Standards Forum XR Accessibility Guidelines.
Jason: background, received request from metaverse standard forums on XR guidelines. Suspect this will be a coordination exercise. Possibly APA and A11y guidelines working group and probably treat as such. Dylan Fox...Looking for a pathway to having this influence some normative document and if going to be normative outcome then i assume a11y
guidelines working group. not really this task force or APA.
Janina: consumptin side, what constitutes a11y and web content that's their role. But that's as far as they go., If w3c writes normative guidance we want to influence that. Agree we can have that conversation if anything on that horizon. XR was a hotter topic 4-5 years ao than now. But could come back with AI.
Jason: question, consider metaverse standards doc, they've done a gap analysis on what they think is missing in a11y fguidelines draft and next version that they'd want to see filled in. One step that could be taken. Another step could be a w3c submission process. Need a working group?
Janina: could be a submission process and probably is. They could submit as is and publish as a statement (not capital S) - we want the world to know this stuff. Everyone wants a working group note. I don't know if they want to pursue that or if we want to. AI, gaming. This would be no higher than third. But, they've worked it up, Not going to say
no. But need to negotiate the when and the how. But there is a place you can publish a working group note of a w3c member.
Jason: not something that's widely cited
Janina: right, no consensus and someone wanting to get opinions note
Jason: personal view, probably a manageable number of additions to a web content a11y guideines that cover gaps made conspicuous by XR technologies. Some point in process can be added in. But needs someone to work out what they are and reduce to what the principles that should be added to the standards. And docs would be written accordingly. Right
way to make anythign normative as far as w3c concerned. Fair amount of work to do it. Some kind of gap analysis with accessibility guidelines. Could be wrong, but that's what seems to be. Just another web content and in accessiblity guidelines purview.
John: lots of people on the list who weren't regularly at meetings. Lots of people from Google. Not just Dylan.
Janina: i'll take it up with APA and send Dylan a note to ask what he wants to achieve inside w3c.
Jason: Sounds like the way to proceed.
Janina: if we continue the relationship, APA would kick it back here. I think we know where this would fit in our priorities.
Jason: Janina, let us know from the APA side
… any final questions on this?
Game accessibility.
Jason: i know this is ongoing topic and development. One comment to add from last week. But first John - anythign to raise to the group?
John: still adding to gap analysis. want to get there sooner rather than later, but competing work priorities. Discussion emails if better to have accessibility features that break the challenge of the game but still allow them in or better to say if a feature that breaks the game it should be in? That's doing the work - getting ahead of yourself.
I think taht's interesting but not for discussion yet
janina: we'll work on that and discuss when we get to the paper.
Jason: one of the capabilities in educational domain, might be relevant here, is that there's...the more sophisticated education systems is adaptive. large database of questions it can ask you. based on response to the first several questions can determine which questions to ask. Dynamic measuer difficulty by your responses. Games might be able to
operate similarly in principle. Can set level of difficulty....not sure if there are games taht already build this in?
Janina: isn't that doing the work? That's a gap worthy of discussion? But not the proposal to doing the work? Goes on bucket list of things to talk about
Jason: yes. We're always interested in accessible systems that adapt to needs of the users. Adaptations that touch on dsireable and undesireable issues.
John: what is intended use of the document? I want to use outcomes of this as a way to get towards regulation, a reasonableness clause in legislations/regulation. Reasonableness argument rather than saying we ought use this doc on how to make any game accessible. Could have different sections in the doc. Temptin to do the work before we have the
permission.
janina: perfectly happy to make hints to regulators. How society on how to reasonably consider how to make decisions and what to expect. What can be done technologically and governance and regulations. And outside the scope of the document. But someone needs to do it
Jason: Most able to achieve in writing about these design questions is giving people concepts and examples and suggestions to help them think about issues more clearly when making decisions. No clear and hard lines. These are ways of thinking about it that we think are useful.
Janina: Judy Brewer was really good at this. Making our pubs had this without saying we're writing this for regulators.
Jason: agree. A clear statement of relevant considersations are and how to approach with creativity and understanding is themost we can expect.
Janina: we can expect judy's help on this when direction is clear and aspect...we can ask her to take a read. She will give us her thoughts.
Jason: getting gap analysis abstract will get us to next stage with APA. Circle back and get into substance and what we want to say in how we make these challenging design decisions.
Miscellaneous topics.
Jason: anything else going on?
Janina: Scott is no longer able to attend Zoom meetings regularly.
… I'd like to leave the door open. I think we welcome his asynchronous work. Very much in keeping with w3c ethos.
Jason: going to be in Australia, So will be in that eastern time for a while. Challenging time zone. Tentatively looking at OCtober and first couple of weeks of November. Still being worked out.
Janina: if we need coverage, I might take that up
Jason: if we can find someone to co-facilitate that would be good to take that aspect up from Scott.
Janina: I'm traveling this weekend but will still be involved. Editing ability constrained by the technology I can bring with me.