I find this ridiculous: mathematicians "ask them to stop testing advanced mathematical problems on proprietary models." I have bad news for them: whatever looks "advanced" math today will look utterly primitive to tomorrow's AI
Timothy Gowers @wtgowers@wtgowers
Sep 29
The advisory group on mathematics and artificial intelligence has just published a set of recommendations concerning the responsible release of mathematical results generated by AI companies using internal models. 1/3 agmai.org/general-sep29/

Sep 30, 2026 · 4:02 AM UTC

26
30
469
40,472
Sort replies: Relevant Recent Liked
Replying to @ChrSzegedy
These are just transition pains. I don't think people have realised society will need to restructure.
1
15
1,747
Obviously. These asks are both impossible and unproductive.
1
1
19
1,519
more replies
Replying to @ChrSzegedy
This starting is just to appease the community, but the rest of the text is good
1
4
477
Yes, the rest looks quite reasonable
4
357
Replying to @ChrSzegedy
The question is basic -> how do you validate the correctness of elaborate proof.
1
3
1,356
Why is that the question here?
1
5
1,260
more replies
Replying to @ChrSzegedy
Someone else notes as much, but I think this is a somewhat disingenuous excerpt They just say that by way of setting up context for quite reasonable and well-reasoned recommendations
1
63
Bad context at the very start. Most of the recommendations sound good.
1
83
Replying to @ChrSzegedy
Excluding proprietary models doesn't prevent them from using AI
1
150
Replying to @ChrSzegedy
This is really insane. The title itself is ridiculous. This should be taken as example of deeply unethical reaction to AI.
4
469
Replying to @ChrSzegedy
It is so silly too. So rich universities should stop doing research in advanced chemistry labs because a poor university in Nigeria can't afford the same?
2
508
Replying to @ChrSzegedy
It's as if software engineers would ask them to stop generating advanced algorithms on proprietary models. Deeply embarassing.
1
193
Replying to @ChrSzegedy
The real question is why the fuck should we not solved these math problems?
1
394
Replying to @ChrSzegedy
Like asking mathematicians to stop developing frontier level mathematics with their top researchers
1
609
Replying to @ChrSzegedy
Ah yes the future fantasy machine we are trying to build with our current primitive mathematics.
1
130
Replying to @ChrSzegedy
It's cringe to see how mathematicians (supposedly smart people !) are making fools of themselves as the modern Luddites.
1
217
Replying to @ChrSzegedy
they just want a robots.txt for their problems
71
Replying to @ChrSzegedy
Then share the prompts, full cot and weights
109
Replying to @ChrSzegedy
AI cannot develop complex mathematical theory, it (cleverly) brute force its way to a semantically sound proofs. That all. He has no concept of advance, primitive, or math at all.
1
156
Replying to @ChrSzegedy
Cody Johnson
@codercody
Sep 25
There are going to be 1,000 new millennium problems
38
Replying to @ChrSzegedy
>Stop lording your unfair intelligence advantage over us! What a call for open AI ~
2
90
Replying to @ChrSzegedy
The response of TCS and math to AI advances is reminiscent of how we as Society responded to Covid in the early days. So in some sense, it's hard to fault them. Exponential curves are perhaps hard to internalize for everyone.
2
158
Replying to @ChrSzegedy
Funny how irrational mathematicians have become.
1
29
Replying to @ChrSzegedy
they are cooked. we have more urgent problems.
1
103
Replying to @ChrSzegedy
ASM@ASM65617010
Sep 30
"what will happen is academia will be made to fit into the AI age"
23
Replying to @ChrSzegedy
On the other hand, we do need some measures against shockwaves of bullshit on @X. Please see glacier-studio.com/blog/inde…
228
Replying to @ChrSzegedy
Its gonna be more awkward eventually because smarter ai will separate accidental complexity from fundamental complexity at some point. Not now obviously as 13M lines of lean shows. That will invalidate many “open problems”.
557