Gill Verdon

@gillverd

Founder & CEO @extropic

following
followers

Claude is said to have devised reversible divsteps without storing branch history

A user quotes an account claiming a 12-fold reduction in decoder gates relative to Google's previous construction.

6 Sources, 3d ago, updated 3d ago

A joke about marketing single-pbits

A user jokes that they wish they'd thought of the single-pbit marketing angle themselves.

2 Sources, 6d ago, first seen 6d ago

A ‘hill issue’ joke moves from San Francisco walks to RL post-training runs

One post jokes about hating walks in San Francisco; another quotes it and swaps in RL post-training runs.

1 Source, 11d ago, first seen 11d ago

Could unified field theory or quantum gravity be solved by October 2027?

In reply, a user said it wouldn’t surprise them much, recalling a past view that focused effort might solve it quickly.

J⧉

5 Sources, 14d ago, updated 14d ago

OpenAI releases results from its internal AI math model on GitHub

OpenAI says it drew on an independent mathematics advisory group’s advice about how to release the results.

435 Sources, 9d ago, updated 9d ago

Extropic uses reinforcement learning to post-train models for Thermo AI research

Extropic says the work aims to speed algorithmic discovery for what it calls a new species of computer.

16 Sources, 17d ago, updated 17d ago

Google introduces Gemini 4 Argon with limited cyber rollout

Google announces a 1 million-token output limit, up from 64,000, for its new model. Initial access goes to trusted cyber defenders through the Fairwind Program.

333 Sources, 15d ago, updated 15d ago

AMD agrees to acquire World Labs in $8.2 billion all-stock deal

World Labs says Fei-Fei Li will join AMD as executive vice president and chief scientist, working directly with CEO Lisa Su. The company says the move will bring its spatial AI research closer to the hardware.

93 Sources, 20d ago, updated 20d ago

A joke about post-training a model to beat GPT-5.6

A post imagines how someone acts after getting a Chinese open base model to beat GPT-5.6 on a niche task.

2 Sources, 33d ago, first seen 33d ago

Waymo’s Ojai vehicles reportedly ride jerkier than its Jaguars

The user urges Waymo to use reinforcement learning to make the Ojai vehicles less jerky.

1 Source, 37d ago, first seen 37d ago

Do AI extinction warnings help or hurt safety efforts?

CBS Mornings reports that current and former Anthropic employees warned AI could kill all humans. Critics worry exaggerated forecasts could undermine trust, while a defender of numerical forecasts argues they convey the seriousness of the concern.