- From: Nayan Padrai <nayan@ecomback.com>
- Date: Wed, 12 Mar 2025 07:48:22 -0700
- To: Gregg Vanderheiden RTF <gregg@raisingthefloor.org>
- Cc: Todd Libby <toddlibby@protonmail.com>, Jennifer Strickland <jstrickland@mitre.org>, WCAG <w3c-wai-gl@w3.org>, Wendy Reid <wendy.reid@rakuten.com>, "Kenneth G. Franqueiro" <kfranqueiro@w3.org>
- Message-Id: <1EC5C634-BAFD-4496-AEA5-876C8008D1DA@ecomback.com>
+1 to Gregg’s comments. Thanks Nayan On Mar 12, 2025, at 7:38 AM, Gregg Vanderheiden RTF <gregg@raisingthefloor.org> wrote: 1 this topic was raised - to help out those who take notes 2 the notes we currently take usually include a fraction of what is said or meant, are often so brief they are not understandable, mostly do not capture even key ideas people present, and often say the opposite of what people intend. My comment was that we should periodically - have an AI take notes (or correctable notes per below) and then compare them to the human notes. Wnen the AI notes ARE more accurate, more complete, and capture key aspects — then we should relieve notetakers from the task and let them participate. If AI iw worse - we can wait and give it a go again in 6 months or a year. I will bet a large amount of money they will be more accurate and complete fairly quickly — and we should save the poor notetakers the task and let them participate when they do. But am happy to be proven wrong if I am. Always like to new data. Remember that with what I prospoesd a person acn look at the summary of what they said seconds after saying it and correct it if wrong. This is how all the AI vs Human arguments should go. With data rather than either overly positive or negative comments - unless the comments help shape the question or the possible solutions. Best Gregg (and may the best note - taker win) OH - and we have one more thing - we should also see which is more accessible as we go. No idea with would be but one more thing to include in our evaluation. > On Mar 12, 2025, at 2:42 AM, Todd Libby <toddlibby@protonmail.com> wrote: > > I see more than enough of AI on LinkedIn and we don’t have to go down the road every tech company is going down these days. Do we? > > I am a staunch opponent of using AI especially in scribing. Why take out one of the best ways that a new person that joins the W3C can start to interact? > > Can AI understand the contextual meaning of what someone has said that may need a human scribe to hear and to detail as much as possible or in the notes? > > What about verbal cues or subtle forms of communication that different people are trying to communicate with? > > Keep the AI out of the meetings and let people participate by scribing if they don’t want to jump right in immediately and would prefer to choose to help out and be a silent participant instead. > > Personally, AI scribing is a horrible idea as is AI period. > > > > On Tue, Mar 11, 2025 at 07:09, Jennifer Strickland <jstrickland@mitre.org <mailto:On Tue, Mar 11, 2025 at 07:09, Jennifer Strickland <<a href=>> wrote: >> >> Hi all, >> >> >> I appreciate your enthusiasm on the topic. I’d like to remind everyone that @Reid, Wendy <mailto:wendy.reid@rakuten.com> is doing the research and already has covered all what is covered here and more regarding AI transcription for minutes for the W3C more broadly. There are other topics we need to focus on right now. >> >> >> Additionally, AI transcription does well on some non-English languages for the record (i.e., Chinese, both Mandarin and Cantonese) and referring to “foreign languages” centers English. We have a lot of different people among our members, even if we don’t encounter them during our Europe and the Americas time zone meetings. >> >> >> Kindly, >> Jen >> >> >> >> >> From: Kenneth G. Franqueiro <kfranqueiro@w3.org> >> Date: Tuesday, March 11, 2025 at 4:49 AM >> To: w3c-wai-gl@w3. org <w3c-wai-gl@w3.org> >> Cc: Gregg Vanderheiden RTF <gregg@raisingthefloor.org> >> Subject: [EXT] Re: RE using AI minutes >> >> RE "it is much more complicated than it first appears": I agree. RE "it would be wonderful for the scribes" to be replaced by AI: I disagree heavily. A couple of your notes mention having a human reviewer or shadow-scribe to an AI scribe. This >> >> >> RE "it is much more complicated than it first appears": I agree. >> >> >> >> >> >> RE "it would be wonderful for the scribes" to be replaced by AI: I disagree >> >> heavily. >> >> >> >> >> >> A couple of your notes mention having a human reviewer or shadow-scribe to an >> >> AI scribe. This makes the human scribe's task _harder_, not easier. Now the >> >> human needs to constantly scrutinize the text output of the AI and be able to >> >> correct the past, while simultaneously keeping up with the present verbal >> >> conversation - and hopefully still have the capacity to participate if they >> >> have anything to add. >> >> >> >> >> >> I consider myself decent at transcribing but recognize that multitasking while >> >> scribing is challenging, and I would absolutely rather do the work myself than >> >> constantly be tracking past, present, and future in both written and spoken >> >> word at the same time. >> >> >> >> >> >> --Ken >> >> >> >> >> >> On Tuesday, March 11, 2025 2:26:04 AM EDT Gregg Vanderheiden RTF wrote: >> >> > The topic of AI came up in F2F >> >> > >> >> > to keep from using up too much time there here are some thoughts on ideas >> >> > >> >> > >> >> > 1) A good way to handle foreign language speaker not being as well >> >> recognized >> >> > - use AI to transcribe >> >> > - have “backup” scribe that only steps in when recognizer faile >> >> > >> >> > 2) for Off minute comments you can turn off the mike to the AI transcriber off >> >> > (see note below - where we need a special AI tool not just the one in zoom >> >> - >> >> > for a number of reasons. >> >> > >> >> > 3) If we use AI summaries -we NEED to implement it in a fashion that >> >> provides >> >> > a method for “in meeting” instant correction of mistakes. In fact we need >> >> > that for the human done minutes which also often misunderstand and sometimes >> >> > scribe exactly the opposite of what the speaker said >> >> > >> >> > >> >> > 4) Remember the distinction between transcript and minutes >> >> > AI can do both but tends to do the summaries only after a meeting. This is >> >> > too late for someone to notice that the summary is not complete - is >> >> > inaccurate - misinterpreted - or is opposite of what was said Suggestion >> >> > Need a special tool to change the transcript into summaries on a per >> >> commenter >> >> > basis - like a scribe does that summarization would go into the IRC >> >> > BETTER though would be for these summarizations (minutes) to appear in a >> >> > separate document that is group correctable so that the commenter can >> >> correct >> >> > the AI summary. per #3 above >> >> > >> >> > 5) Speaker ID >> >> > This is often cited as an AI shortcoming. But it is just as much a problem >> >> in >> >> > human minutes. our best solution to this for both is using queue and having >> >> > chairs announce each new speaker (as they do) Sometimes things get into a >> >> > back and forth between two people >> >> > Best solution is for each person to say their name before the speak each >> >> time >> >> > ALSO >> >> > If we have a human “shadow scriber” that is just monitoring the AI - they >> >> > could add speaker names where the AI does not. >> >> > >> >> > 6) for hybrid meetings - I suggest we purchase more microphones so there is >> >> > no-one >> >> > >> >> > >> >> > IF WE WERE TO USE AI — given all the above and the comments I think we >> >> > should build (or find) one that does what we need. When you think about it - >> >> > it is much more complicated than it first appears to replace or even augment >> >> > the human scribe. But it would be wonderful for the scribes once you do. >> >> > >> >> > Gregg >> >> >> >> >> >> >> >> >> >> >> >> >> >> >> >> >> >> >> >> >>
Received on Wednesday, 12 March 2025 14:48:39 UTC