The glazer guy blocked me so I can't reply. My answer is it's not about dick measuring but preserving the basic standard that you ought to only release something you understand. What's the gain from releasing a result a few month earlier that makes it worth killing this standard?
Abuod retweeted
If it's not clear to the commonfolk by now, it should be obvious that multiple decades of work by dozens of mathematicians on a topic that is only directionally bound to an open conjecture is order of magnitude richer in content and is far-reaching beyond the said conjecture than an AI generated shortcut proof of the conjecture with a Lean certificate.
But sure, go ahead and defund it all because a computer can solve all open problems, I don't care. If society wants to strangle scientific culture, and by extension itself, in the backalley, then I don't particularly mind.
Abuod retweeted
increasingly concerned that ai is revealing the mathematical world as negatively curved. instead of everyone wandering about in idea-space and clustering around the same famous conjectures (local positive curvature), ai is pushing out the boundaries rapidly and revealing that the space of mathematical ideas is ~hyperbolic (negative curvature). though this means there's lots left to discover, it also means that small separations between people/agents will lead to wandering alone.
we may be emerging into an idea-space where there's so much to see and do but everyone is alone in a private heaven, unable to share their subtle joys with others...
Abuod retweeted
from the latest preprint on analytic prismatization by anschütz, le bras, rodríguez camargo, and scholze:
Abuod retweeted
Maybe you need to love math unconditionally to stand the disruption. Nobody has promised anything to anyone.
The situation that Ramanujan was through doing math as a clerk in Madras' port, or Galois when he was detained in Sainte Pélagie prison, was much worse than the situation of any contemporary mathematician can face in the AI era. They didn't give up. They kept filling their notebooks and brouillons with their mathematical dreams. They weren't doing this to get a position, status or a way of living. They were doing it for the love of math and to be remembered by History.
I can only talk for myself. My personal goal is to discover new unknown mathematical landscapes that are behind these mountains we have in front. For others their goal may be different. Their challenge is to climb these mountains with they effort and sweat. Now, we do have an elevator. I understand why they are unhappy. I also miss the challenge and sport side of climbing. For the highest mountains there are still no elevators. With the new elevators I can see many new landscapes of the future. Something that I never dreamed to see in my lifetime.
Do you understand why I can feel happy?
At the same time I am sincerely sorry for those that do not share my excitement. I feel sorry for them for their pain. I feel sorry for them in the same way that I feel sorry for those that never knew true love.
Too true:
“I am mathematics PhD student. Up until recently, I fully intended to pursue a career in academic research. I deeply love mathematics, and I entered graduate school because of this. However I have made the decision to leave academia once I finish my doctorate and I am heartbroken; I would like to explain why.
This decision is not because of job security, nor is it because AI can do better mathematics than I ever could. Rather, I am becoming increasingly convinced that this profession is morphing into something unrecognizable, something alien, something I did not sign up for.”
Abuod retweeted
“The ultimate hidden truth of the world is that it is something we make, and could just as easily make differently”
— David Graeber (1961-2020)
Link: tinyurl.com/y9a24tv9
Abuod retweeted
"If we want to continue progress in science, we must be prepared to become religious."
This isn't something new when you think about it. Most jobs are "bullshit jobs" today, in the sense that everyone is a "manager". Even in research, most seniors don't do any research, they just manage their army of PhD students. Now even juniors are managers with AI.
Funny how deep learning theory research is all useless while deep learning empirical research is all bullshit
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com/
The Temptation of The Three Crosses
The Temptation of The Three Crosses II: The Revenge of The Twenty-One Savages
The Temptation of The Three Crosses III: The Return of The Youthful Bandit
