Hacker News: vatsachak

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Thu, 21 May 2026 01:37:58 +0000

It might just be that we didn't have enough compute till now. StockFish definitely has superior intuition

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Thu, 21 May 2026 01:36:47 +0000

They are most likely using them in training. I doubt their IMO team are show ponies

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Thu, 21 May 2026 01:34:47 +0000

They most definitely threw in rl with formal verification somewhere between GPT 4 and now. The models are better at not hallucinating. I don't think their IMO team are only show ponies...

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:53:21 +0000

My claim is that we haven't even witnessed the move 37 of math yet. I am claiming that math AI is going to get even better

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:52:38 +0000

Fair. Let's see in a year. I'm willing to bet that nothing happens.

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:50:55 +0000

Math is a very specialized subset of art and science more amenable to automation.

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:49:19 +0000

My claim is that LLMs waste a lot of time training on all available data.

Math is a sequence of formal rules applied to construct a proof tree. Therefore an AI trained on these rules could be far more efficient, and search far deeper into proof space

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:47:13 +0000

Well, for starters AI doesn't have goals. If there was a super intelligence with goals, why would they work for us?

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:44:58 +0000

Not necessarily. Obviously playing Kasparov on the board requires more planning ability than managing a McDonald's but look at where chess bots are now.

There's much more to being human than our "cognitive abilities"

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:42:53 +0000

Yeah, but I wouldn't be surprised if they train the model on verification assisted by Lean.

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 20:42:44 +0000

Yeah, but I wouldn't be surprised if they train the model on verification assisted by Lean

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 19:32:34 +0000

I absolutely believe that AI will supercharge science.

I do not believe it will replace humans.

New comment by vatsachak in "An OpenAI model has disproved a central conjecture in discrete geometry"

vatsachak — Wed, 20 May 2026 19:30:59 +0000

As I have stated before, AI will win a fields medal before it can manage a McDonald's

A difficult part was constructing a chess board on which to play math (Lean). Now it's just pattern recognition and computation.

LLMs are just the beginning, we'll see more specialized math AI resembling StockFish soon.

New comment by vatsachak in "Google I/O"

vatsachak — Tue, 19 May 2026 17:57:33 +0000

Powerful. AI

New comment by vatsachak in "War game exposed U.S. vulnerability to low-tech warfare"

vatsachak — Tue, 19 May 2026 05:38:22 +0000

Everything in this country can afford to be a grift if everyone treats the treasury as an infinite source of capital.

This is a systematic issue which very few people care about.

New comment by vatsachak in "Anthropic acquires Stainless"

vatsachak — Tue, 19 May 2026 05:31:11 +0000

Powerful AI is here as Dario said in 2024. Open your eyes. Anthropic acquired Stainless because they know how to use Claude better than their own employees

New comment by vatsachak in "Where Are the Vibecoded Photoshops?"

vatsachak — Mon, 18 May 2026 14:52:35 +0000

Sure. Where is the vibecoded Figma or Lightroom?

New comment by vatsachak in "A message from President Kornbluth about funding and the talent pipeline"

vatsachak — Thu, 14 May 2026 17:20:33 +0000

I have solved open problems of fields medalists and can't get a job in academia. I currently make 4800 a month after taxes as a lecturer in San Diego, pivoting to SWE. Math PhDs are having a hard time

New comment by vatsachak in "Maryland citizens hit with $2B power grid upgrade for out-of-state AI"

vatsachak — Mon, 11 May 2026 02:37:14 +0000

They're chronically online because millennials basically shoved digital heroin into their face from birth

New comment by vatsachak in "Meta's embrace of A.I. is making its employees miserable"

vatsachak — Sat, 09 May 2026 22:33:13 +0000

Technology can also divide power. Think about the amount of open source intelligence that exists.