Stéphane BortzmeyerGood morning, Dublin! Seventh and last day of #IETF121 https://www.ietf.org/meeting/121/
Today, for me, RASP meeting (auto-study of IETF).
We'll continue to go deeper in details:
OK, start of the proposed research group (like a working group but for the #IRTF) RASP https://datatracker.ietf.org/meeting/121/materials/agenda-121-rasprg-04
Justus Baron present on the geography of standardization.
Role of the "global south" in standard setting, etc, remote vs. physical meetings.
Distance and borders matter!
The speaker worked on IETF and 3GPP, with data from meeting attendance..
Gender participation statistics: the speaker's research use the first name to infer gender, which is brittle.
(IETF does ask the participants their gender. But it is no made available to researchers.)
Now, a talk about generating automatically reports of IETF meetings (in LaTeX!) with AI…
May be this live-tooting will be done by an AI at the next IETF meeting?
"It sometimes takes a creative detour" (= hallucination).
End of #IETF121 and end of this long live-tooting. Fly safe back home and have a beer.
Albert ARIBAUD Ⓜ@bortzmeyer I just don't get that use case. Either a talk is not worth reporting on accurately, and then it's not worth raporting on at all, or it's worth reporting on, and then inaccuracies "by design" are just unacceptable.
@aaribaud I guess the idea is that the unreliable report will at least give the readers an idea of what was discussed, often the most requsted info in IETF reports. Once you know that we had a big discussion about composite PQ/T keys in the OpenPGP group, you can search in depth, even if the report wrongly explained the issue.
@bortzmeyer Except, if the report explains the issue (whether correctly or not), people *won't* look further; they'll assume the report explains it right, because who in their right mind would publish a report with errors in it?