vatsachak

5 ms·

1.
by vatsachak 3d ago
They actually don't
2.
by vatsachak 3d ago
The first time I have agreed with David Sacks
3.
by vatsachak 3d ago
Yeah I use Astra Light at work. It's great but it probably cannot work on a project longer than a week without introducing architectural blunders
4.
by vatsachak 4d ago
Slop posting
5.
by vatsachak 4d ago
Paper written by AI
6.
by vatsachak 4d ago
Yeah. Not to mention the paper-clip maximizing that RL induces in them. I had 5.6-Luna use a parser combinator lib in order to find out that it imported it but wrote its own parser, so it technically followed my instructions. Imagine that p
7.
by vatsachak 4d ago
He's insinuating that AI will end the world lol. Nothing ever happens
8.
by vatsachak 4d ago
Nah the AI won't replace coders or mathematicians until it can maintain codebases/knowledge long term. I actually don't think that current AI can replace any profession that requires human interaction over many weeks. This is
9.
by vatsachak 4d ago
Why would China or anybody else co-operate unless they have access to OpenAI levels of compute and success with training? This sounds like pre-IPO hype. They better just try and top astra.
10.
by vatsachak 5d ago
Who cares about the prize and the outdated methods? OpenAI and Anthropic might have 3 millennium problems by December
11.
by vatsachak 5d ago
In no way, it's just a hobby.
12.
by vatsachak 5d ago
Maintainability is hard because it requires much more planning than correctness. You need to think about how you can construct code which gives you the most amount of reusability while not compromising on readability and also thinking about
13.
by vatsachak 5d ago
Because programming is fun
14.
by vatsachak 5d ago
What's the threat? I would use it to write a dependent type theory that is JIT compiled and use it to rewrite emacs.
15.
by vatsachak 5d ago
I actually read through a bunch of LLM slop on the weekends trying to catch white whales I couldn't during my math PhD. The incentive is solving the problem and understanding the solution.
16.
by vatsachak 5d ago
Apologies. Noted
17.
by vatsachak 5d ago
That means that academia is outdated not that LLM math is bad
18.
by vatsachak 5d ago
Both are true I admit
19.
by vatsachak 5d ago
How is it plagiarism if no one else had the full proof idea first?
20.
by vatsachak 5d ago
That's fair and I agree. But stockfish can play successful middle games and there's no reason that AI cannot create research programs
21.
by vatsachak 5d ago
No that's not sufficient but that's not what's happening. Why do we believe that we cannot train models which could explain the jargon in more human terms when current LLMs can perfectly explain the most complicated codebases
22.
by vatsachak 5d ago
https://cdn.openai.com/pdf/32d9f210-8b73-45e0-91bc-82a30aef8...
23.
by vatsachak 5d ago
I have a paper in JEMS and have solved 20 year old conjectures. I also know that math academia is a catty game which is mostly driven by politics and ego
24.
by vatsachak 5d ago
Grothendieck accused Deligne of "slop" with his proof of the weil conjecture. Scholze's math is definitely not slop. But you're taking two of the best mathematicians of the last century against my claim about averages
25.
by vatsachak 5d ago
Because it's beautiful. Because you love math for maths sake and not some weird egotistical game
26.
by vatsachak 5d ago
Any serious mathematician would read the LLM output and rework their understanding. I read a bourgain paper a week in grad school and they're probably worse than an LLM generated paper. I still had to recreate the tricks in my own lang
27.
by vatsachak 5d ago
You mean manufacturing
28.
by vatsachak 5d ago
Exactly. Mathematicians are trying to cope that math wasn't slop this whole time. Grothendieck was anti-slop but most papers are slop. I don't think AI is going to rewrite bourbaki anytime soon
29.
by vatsachak 5d ago
[flagged]
30.
by vatsachak 6d ago
The irony of LLMs is that OpenAI provided software is still just a wrapper. The only way they can get more users is through exclusive discounts on their wrappers

More ›