Meeting minutes
Accessibility of Machine Learning and Generative AI.
jasonjgw: Merged Josh's changes into main
jasonjgw: Included resolving merge conflicts
jasonjgw: Believe we're now at deciding next round
jasonjgw: Also question of citation support
Josh: Made a few minor edits following jason's merge, will push as soon as I can login
Bibliographic Citations
jasonjgw: Citation Style Language is a useful XML format supporting many other formats
jasonjgw: Also bibtext
jasonjgw: Suggest this may be easiest
scott: asks about exporting
scott: Asking about export from Endnotes -- the db application for bibliographic references
jasonjgw: I don't have Endnotes, but please check what formats it exports
scott: Can do
jason: i think we have a clear idea what to do, need to determine best tools, and then integrate into document production
scott: endnote - all the good options available
jasonjgw: bibtec and IRS are best options, I think
jasonjgw: scope section, josh's outline that needs to be filled in with content. Anyone interested in looking at any of that?
josh: we'll collaborate on that. placeholder and asking a question in there. But we'll collaborate on that.
jasonjgw: drafted based on discussions in meetings, can go back and draft with v2 with what i think we agreed upon or circling around in discussions
jasonjgw: BibTeX most widely supported
jasonjgw: need something like citation style language of what generated bibliography should look like. Generate the citations and references, done with a tool written for that purpose. Integrate integration of that tool into the document build process. XML citation style that defines what output looks like.
janina: lets worry about what it looks like as it happens. Scan through databases, if working with endnotes...what you give the w3c process is a text file that has records in the bibtex format.
jasonjgw: in document will be labels where citations are that correspond to the names of the records
janina: presuming kevin will find that on their own? something needs to describe thw ASCII text file, can't be so proprietary...
janina: "we prefer that you support bibtex, fallback is RIS, bibtex writes in ascii text file, described here, suggest CSL to transform the data into formats usable in w3c HTML publications...
janina: need a start of authority to tell them how to use them and what's included in the ascii file.
jasonjgw: need to choose a tool that already does this and integrate it.
janina: Raja is on board with all of this? Want to check.
Scott: this all works for me. pandoc would be good too
jasonjgw: i expect to be woprking on purpose and scope doc with Josh and others
janina: fall back name - RIS? Need to have the same logic path, point fallback if Bibtex is problematic to work with, need fallback with the logic path. CSL still relevant at that point.
janina: propose to write the email together in the next couple of days
scott: happy to help with the mail too
document revisions
jasonjgw: taken action to look at purpose and scope. Josh has changes coming shortly editorial in nature. I'll help merge those in. Anyone who wants to take up any aspects in augmented doc that includes outline?
jasonjgw: anything else need to take care of inthe next week?
janina: go thru github issues and take notes in new structure and link actions
janina: that will tell us if we do or don't have appropriate doc structure. Should be an early task
scott: working session on next week's call?
josh: sounds good
jasonjgw: sounds good
janina: someone in the doc making notes on the fly? use the IRC channel...review issues one at a time, drop web pointer in IRC, copy and edit in document (in template)...do it that way?
janina: add notes and commit, need someone to track, if someone can monitor the issues and make sure the link gets into IRC. different process suggestions?
josh: I'll tag team on this
jasonjgw: broader ethics might need to find a place inthe intro with reference to sources. Known issues...
janina: i think sustainability should be in there too
scott: lots of interest as I've been informed. even if there's a line about "not in scope" is strongly recommended
janina: what have we wrought with this web, a lot of people want to know about sustainability. electrical generating capacity to AI farms that need to be built
jasonjgw: sustainability is complicated. Not sure how much addressed in this document? Who's best at making decisions on renewable sources?
scott: not knowing may be the answer at this time. Something to factor in future developments.
jasonjgw: run locally or on a data center, and who's data center?
janina: don't forget accessibility just because trying to be energy efficient. IF build good accessibility, content will be more efficient and effective?
scott: live captioning on local removes latency and presumes to be more efficient? food for thought. might be more there...a few possibilities as we talk this through
janina: maybe it is a section somewhere and not intro?
jasonjgw: it depends on how much to say. Maybe separate subsection in intro? other broad ethical considerations useful as well. ethics of webmachine learning might be useful and main text on a11y considerations. Broader issues in there?
misc
Miscellaneous topics.
janina: interest in discussion among management, asking to resume work on something we started a few years ago - AUR for web authentication. Less to say more about this now than then? There is a wiki page where we started gathering thoughts. Related to CAPTCHA.
janina: larger task of killing off CAPTCHA. Need to be done with this version of CAPTCHA
jasonjgw: prove humanity without revealing identity. use pass keys and FIDO 2 to avoid other mechanisms
janina: need to say a bit more than that. But a lot has been solved from what we would have said 3-4 years ago. point to that and where we are today
jasonjgw: strong motivation to deprecate passwords and move to passkey authentication. all major tech corps behind this. Anything we say to hasten this? Already moving fast.
jasonjgw: maybe treat CAPTCHA and authentication in same doc?
janina: it's on me. want to hear their concerns on why take it up.