Hacker News: sashank_1509

New comment by sashank_1509 in "AI is mass decompiling games. That's both awesome and scary"

sashank_1509 — Wed, 14 Oct 2026 17:56:19 +0000

Yes, games are not furniture. I have no idea wtf if a game is functional even means?

New comment by sashank_1509 in "Linus Torvalds's first LLM-assisted patch series"

sashank_1509 — Wed, 14 Oct 2026 17:54:28 +0000

Srsly

New comment by sashank_1509 in "AI is mass decompiling games. That's both awesome and scary"

sashank_1509 — Wed, 14 Oct 2026 16:41:25 +0000

This is an opinion that is only plausible if you completely devalue video games as an art form.

Will you prompt the movies you want instead of watching a movie made by someone else? Will you prompt the novel you want to read? It’s nonsensical in other art form, it’s nonsensical in video games too unless you go for the lowest common denominator friendship or similar.

New comment by sashank_1509 in "Gemini 4 Argon"

sashank_1509 — Wed, 14 Oct 2026 03:56:29 +0000

Well I work at BigCo and I have no idea how it compares to Claude Code because I’m not allowed to use Anthropic models outside of our custom company harness. Opus 5.5 works extremely well, it can launch multiple subagents, assign them tasks, use our company’s inscrutable build system (that no one really understands anymore), and run tests, all by exposing harness commands in the system prompt. I don’t think we are lacking in any way.

New comment by sashank_1509 in "Gemini 4 Argon"

sashank_1509 — Tue, 13 Oct 2026 23:37:18 +0000

I’m curious at the enterprise level does Anthropic have staying power. My sense is when you use it with your own custom harness, it doesn’t, I can switch between Opus 5.5 or Gemini Argon and not really care as long as capabilities are similar. This is good for the consumer, and bad for the model developers.

New comment by sashank_1509 in "500B tokens later: Letting AI agents decompile a first-person shooter"

sashank_1509 — Mon, 12 Oct 2026 04:35:41 +0000

Are they replicating physics and the rendering pipeline, or are they basically using known entities like Unreal, Havoc etc but letting the agent tune it to match the game?

New comment by sashank_1509 in "500B tokens later: Letting AI agents decompile a first-person shooter"

sashank_1509 — Mon, 12 Oct 2026 02:49:21 +0000

What is the level of a game that can be decompiled this way. Can a game like say assassins creed or Crisis be decompiled this way? How are the art assets replicated (do we provide it separately). It seems crazy if this is possible!

New comment by sashank_1509 in "Why Texas Is Making Data Centers Wait"

sashank_1509 — Sun, 11 Oct 2026 17:10:38 +0000

I don’t want to be hyperbolic, but to me this technological revolution more than any other in history feels like Silicon Valley and its VC class going to war with rest of human society.

> AI is needed to reindustrialize America.

It seems like they have no idea or are willfully deceiving why Americans want to reindustrialize USA. It’s not because they feel bad that things aren’t made in USA anymore, it’s because they want those manufacturing jobs back. If you replace humans with robots, it’s almost like a monkey paw’s fulfillment of a wish. They want to build 2 Trillion of data centers to get rid of your white collar jobs and your blue collar job, and you’ll be happy for it cause GDP go up. SMH

New comment by sashank_1509 in "Vibe coded browser ports of Halo, The Simpsons: Hit And Run, GTA work well"

sashank_1509 — Sat, 10 Oct 2026 19:14:20 +0000

I don’t know why paying AI companies is better than paying devs for their actual work. Pretty sure this won’t work though outside of the most cookie cutter Unreal Engine games etc.

New comment by sashank_1509 in "Sharing AI progress in mathematics"

sashank_1509 — Thu, 08 Oct 2026 20:12:41 +0000

You do you but I think that’s cope.

New comment by sashank_1509 in "Trump administration is suspending Microsoft from a green card program"

sashank_1509 — Thu, 08 Oct 2026 19:13:09 +0000

Take this problem : “cheap outsourcing of simple coding work”, and replace outsourcing with AI, which will be far cheaper, much faster and much more compliant than any human being, and you can just imagine the disastrous situation we’re sleepwalking into.

To all the idiots who claim AI will create new jobs, will only change the work humans do, why didn’t that apply to Indian outsourcing?

As someone who uses Opus 5.5 daily, I feel like this admin is trying to put out a small fire, while completely blind to the all encompassing forest fire just around the corner.

New comment by sashank_1509 in "OpenAI withdraws three mathematical results"

sashank_1509 — Thu, 08 Oct 2026 15:25:42 +0000

Also by my understanding it’s not an algorithm, it’s just an upper bound. Some other model (presumably) needs to find the actual algorithm now

New comment by sashank_1509 in "OpenAI withdraws three mathematical results"

sashank_1509 — Thu, 08 Oct 2026 15:21:37 +0000

These errors were discovered by human mathematicians

New comment by sashank_1509 in "OpenAI withdraws three mathematical results"

sashank_1509 — Thu, 08 Oct 2026 15:20:45 +0000

An actual mathematician found the errors based on my Twitter feed. Once the error is pointed out GPT Astra also confirms that it’s unfixable.

New comment by sashank_1509 in "OpenAI withdraws three mathematical results"

sashank_1509 — Thu, 08 Oct 2026 15:18:52 +0000

Super-intelligence that’s going to end mathematics as we know it surely must be held to a higher standard than puny humans?

More seriously the problem is the complete utter lack of care OpenAI has shown in their desperation to demoralize mathematicians with their new LLM. In their words, it took 3 hours of ChatGPT pro per result, why not spend a hundred hours per result formalizing it, checking if the formalization matches the natural language proof, and whether the argument could be made more simpler and readable. Any human paper has hundreds of hours of work put into it, but OpenAI who is absolutely adamant in demoralizing the mathematical community and demonstrating their superior “intelligence” will only spend 3 hours, write unreadable, inscrutable proofs, not formalize all of them, and then dump it on the mathematical community for some reason.

New comment by sashank_1509 in "OpenAI withdraws three mathematical results"

sashank_1509 — Thu, 08 Oct 2026 07:08:50 +0000

Early sentiments are a lot of the write ups still read like slop and it feels very rushed and not very polished.

New comment by sashank_1509 in "A 100x faster* alternative to homebrew"

sashank_1509 — Thu, 08 Oct 2026 07:07:03 +0000

I guess the LLM reward hacked instead of solving the problem

OpenAI withdraws three mathematical results

sashank_1509 — Thu, 08 Oct 2026 07:05:40 +0000

https://github.com/openai/math/blob/main/history.md

Comments URL: https://news.ycombinator.com/item?id=50002650

Points: 381

# Comments: 612

New comment by sashank_1509 in "Sharing AI progress in mathematics"

sashank_1509 — Thu, 08 Oct 2026 00:12:36 +0000

That only works for art where the customer doesn’t know what he wants. Most people don’t work on Video-Games.

New comment by sashank_1509 in "Sharing AI progress in mathematics"

sashank_1509 — Wed, 07 Oct 2026 19:14:58 +0000

I disagree, there’s no future in “Steering” a model. You ask what you want, and the model delivers. Opus 5.5 is getting there already, A single line prompt creates a fully functional game engine that can be used to make many different games : https://m.youtube.com/watch?v=R_uf5OfMGio&t=3555s (Opus 5.5)

Please explain to me what great prompting / steering you will do when your customer can just prompt exactly what he wants and get a product even more tailored to his needs.