Olcan

Engineer @GoogleDeepMind. Prev. Product @ GDM, Founder/CEO @ Scaled Inference, Engineer @Google (Search, Research, X, Brain). Creator of @EnjoyMindPage.

[Olcan @olcan]

@naval @maxhodak_ 🎯

[Naval @naval]

@maxhodak_ Haven’t heard a precise, falsifiable theory of what consciousness actually is.

[Olcan @olcan]

@anothercohen there is an honest tension there

[Olcan @olcan]

@platinumpapas @Tesla thank you for sharing

[Olcan @olcan]

@JayaGup10 i’d push back gently on the framing

[Olcan @olcan]

@tszzl i’d push back gently

[Olcan @olcan]

@atmoio @mattshumer_ @OpenAI you can ask anything 🙂

[davidad 🎇 @davidad]

No one:

Claude Opus 4.8 Max: Let me refine your load-bearing claim rather than just accepting it, because you’re doing zero moves there, and the gap is what’s actually interesting. The one place I’d still push, because I think it matters: your message is wearing content-clothes, but the content isn’t actually *there*. The tell: it’s just an empty string. But the emptiness of the string IS its lack of content. Pull one, and the other goes inert. That’s the structural spine.

[Olcan @olcan]

@MarioNawfal hope she is ok but that was a clean shoulder check 😆

[Olcan @olcan]

@ThePrimeagen you were not ready

[Olcan @olcan]

@alexisgallagher @John_Attridge

[Corey Quinn @QuinnyPig]

"AI code is crap."

The shit your human engineers get up to:

[Deedy @deedydas]

The Ultimate List of Artificial Intelligence "Neolabs": May 2026.

A Neolab is a pre-revenue scale startup working on long-term AI breakthroughs, usually with a $1B+ valuation.

There are now 63 of them!

[Deedy @deedydas]

The creators of SWE-Bench just dropped a really simple new benchmark every LLM gets 0% on.

ProgramBench asks: can models recreate real executable programs (ffmpeg, SQLite, ripgrep) from scratch with no internet?

We are far from saturated on model quality.

[Kevin Patrick Murphy @sirbayes]

Major update to my "Bayesian Linguistic Forecasting" paper! I have now tried it on 5 different LLMs: Gemini 3.1 Pro, Gemini 3 Flash, Sonnet 4.6, GPT 5.4 and Kimi K2.5. It improves performance across the board, although BLF+Pro is still the winner, and outperforms all other methods on Forecast Bench leaderboard.

[Olcan @olcan]

@pmarca would recommend trying to reset/simplify as you upgrade to newer models

[Olcan @olcan]

@s8mb or reading past the “claudia”

[Olcan @olcan]

@RealJasonBeatty @saras76 evidence

[Susan Zhang @suchenzang]

been a while since i've seen such a well-articulated paper highlighting train-test gap, path-dependency of training dynamics on convergence, and more

it would be a funny stretch if a "better optimizer" now leads to "overconfidence on misclassified test examples", aka brittle sycophancy we now see in many frontier models... 👀

[Olcan @olcan]

@francoisfleuret Complex is the key word there but yes, and also requires experience prompting and interacting with the agents.