2023.03.30.

[bbeatrix]

Dear All,

We plan to have journal club this week as well on Thursday from 16:00 in person at ELTE 5.128 and online on this meet link: https://meet.google.com/fvd-soiu-ent. If you would like to present something this week on Thursday's jurnal club, please let us know by commenting on this issue. We encourage the students of previous semester as well to come and present some current interesting papers from the discussion/suggestion thread (@Dudgit, @borbende, @gebhaard and others). Last week we had a presentation about NERFs, and a short demo of Github Copilot, we still haven't discussed ChatGPT/other GPT experiences and best practices, so if we do not have other presentation, we can talk about this topic!

Best wishes, Bea

[icsabai]

If nothing else comes up, browsing through some of the 500 pages of the "Sparks of AGI" paper https://arxiv.org/pdf/2303.12712.pdf will be fun and may give useful hints for using GPT.

[bbeatrix]

Related to GPTs: if somebody feels that the GPT and LLM improvements are hard to follow, there is an open letter signed by researchers like Yoshua Bengio, to pause giant Ai experiments and training large models for a certain period, so that everyone can keep up with the AI revolution :)) https://futureoflife.org/open-letter/pause-giant-ai-experiments/

[bbeatrix]

The journal club has started, everyone is welcome to the discussion about GPT and other topics, without paper presentations today.

[icsabai]

Re: @bbeatrix 's post on Yoshua Bengio's open letter.
I agree that this may get dangerous. But most probably this cannot be stopped now even if there would be an agreement. Could we stop alcohol or drugs with bans? Or gain-of-function virology experiments? We may ask chatGPT how to solve this ...