Super happy to see Gemini 3 released! ✨️🥳

And thanks for posting this thread on Bluesky! 🙏💙

+1, following now, thanks @tedunderwood.com 👍

Learn more about how Gemini 3 can help you learn, build and plan anything → goo.gle/4oUEkVu

💯 !

Thanks!

💯 I am thinking it's exactly this.

Specifically, I thought Buffet said:

"If the ratio approaches 200% -- as it did in 1999 and a part of 2000 -- you are playing with fire."

recently, but he didn't. He said this in 2001. 😅

So has he changed his opinion in 24 years... ? Or just looking to "make a quick buck"?

2/2

Yes I have heard similar stories.. that's an example of conference reviewing being broken. No consequences for this AC/SAC, I am guessing.

If we *do 🙄

I mean calling them "guidelines" practically implies zero enforcement out of the gate.

So, sure, I can verbally abuse my reviewers and AC and still get my paper in.

I believe any code of conducts we have are for in person conference behavior. I meant one for reviewing.

Someone else said the same but I don't believe so...? Or, if we don't, they are not strict or descriptive enough because there's no enforcement/consequences AFAICT.

Won't be me if that's what you are saying.

I am taking an indefinite break from conference reviewing after this year's AAMAS.

Hmm.. not sure I would agree that AI safety is imaginary or not as practical. These models are being very widely adopted, and if multiagent/games approaches could improve safety generally, I would say that'd be a practical use of games.

It's the inherent difficult involved. Cicero took a whole team and Meta-level resources. It'll def be cool to see that kind of result again but it's always a big risk and so sometimes hard to justify even if you are willing to put in the hard technical work.

This is a concern I have over the wider use of LLMs. Everything is starting to feel fake.. so much that the smallest dose of authenticity is really noticeable and will become overvalued in the next few years.

I still do, but I admit that I tried once. LLM generated reference letters are either so obviously fake that I feel embarrassed using them or it costs more time providing context that I don't save any time.

But there is a still a lot to learn and understand, and these communities don't always talk to each other nor hang out at the same venues. 😅