Eugene Vinitsky

Out of curiosity I had Fable write a response to the article about the weaknesses of LLMs: www.nytimes.com/2026/06/30/o... It's uh, pretty good. claude.ai/code/artifac...

Opinion | Why A.I. Won’t Steal All Our Jobs - The New York Times nytimes.com

Fantastic paper demonstrating how LLM editing is gradually distorting our writing and the way we think to write: arxiv.org/abs/2603.18161

How LLMs Distort Our Written Language arxiv.org

LLMs will likely be the most powerful propaganda tool that has ever existed arxiv.org/abs/2606.16475

AI systems out-persuade expert humans arxiv.org

This is a really excellent benchmark, pointing to some serious missing capabilities in coding agents: arxiv.org/abs/2603.247.... They don't write code with an eye towards maintenance!

SlopCodeBench: Benchmarking How Coding Agents Degrade Over Long-Horizon Iterative Tasks arxiv.org

New Paper: arxiv.org/abs/2606.19370 Self-play yields capabilities but requires frustrating cost-function tuning. Surprisingly, just 30 minutes of demonstration data produces much more human-like driving policies! Led by @daphne-cornelisse.bsky.social Website: spiced-self-play.com

Human-like autonomy emerges from self-play and a pinch of human data arxiv.org

These are very odd failures! arxiv.org/abs/2211.00241

Adversarial Policies Beat Superhuman Go AIs arxiv.org

Does physical modeling matter or should we just do data-driven stuff? I feel like this paper characterizes things perfectly: arxiv.org/abs/2410.23179. Physics-based modeling is always helpful but the relative gain closes at scale

Does equivariance matter at scale? arxiv.org

Would be interesting to see this offered as a service at conferences and journals...https://arxiv.org/abs/2608.13331

Training AI Scientists to Replicate Research arxiv.org

A lot of my fears about LLM usage come from using Stackoverflow when I was a wee programmer. It was easy to fall into copy pasting behavior that long term made you worse

Repulsed by the idea that we should sit idly by while frontier labs determine the trajectory of the AI transition. They don’t know how to make systems “safe” or “aligned“ nor do any such notions make sense without democratic input. There’s so much research to do here.

My position on AI consciousness is that you really should try to eat less meat if you can

Taking more seriously the claim that recent ML models do not "reason", it still is quite odd the particular ways that superhuman game-playing models fail, suggesting that what they're doing is something quite different from human reasoning about play

Very obviously the outcome of people mobbing those who publicly admit they use AI is everyone continuing to use it but hiding it

"LLMs mean you don't have to know things, you can just do things" assumes a clean separation that probably doesn't exist. You know what you want to do in part because of what you know / understand

I'm not worried about people mistreating AI if it ever seems conscious. Humans have an incredible history of taking good care of conscious beings. After all, 99% of humans are vegetarian.

Proposed norm: you're allowed to respond to an unedited LLM document with an unedited LLM document that is twice as long

(1) Artists are justified in being angry about AI and we really shouldn’t use AI art where otherwise we would have hired an artist. Art isn’t really about productivity growth and automating it helps no one. (2) Programming is different; we are bottlenecked on software and need more of it urgently

I have already met some number of people who are basically a front-end to an LLM