Samuel Hammond argues frontier AI progress is shifting from pretraining compute scale to post-training RL environments

Christopher C. Smith@christophcsmithTECH

@hamandcheese Agreed, but GLM-5.2 gives a strong base to build on, which makes it anyone's game, and RL environments *can* simply be bought; there are lots of people building these.

Samuel Hammond 🦉@hamandcheeseTECH

Pretraining scale originally mattered for fitting the curve of human language at ever higher resolution, but now matters more as neural capacity for sample efficient latent reasoning, long-horizon planning, and other capabilities absorbed through post-training. This makes it harder to simply buy your way to the frontier with compute scale. All the low-hanging webcrawl data have been picked. The data needed for post-training is much closer to a kind of "learning by doing" bootstrapped from tons of diverse reward environments and user-agent OODA loops. This requires running billions, maybe trillions of micro-experiments. The $60b SpaceX / Cursor acquisition illustrates just how lucrative and expensive this sort of data is getting. Outside of model distillation, you can't simply "jump to the end" and zero-to-one a frontier agent from scratch. Like a child growing into an adult, you have to climb the developmental ladder rung by rung. This creates a compounding advantage to the companies already in the lead, and helps explain why Meta has struggled to catch-up in spite of their impressive compute stockpile. And as we inch closer to RSI, these incumbent advantages only intensify.

Herbie Bradley@herbiebradleyTECH

@hamandcheese but you can still buy your way to the frontier, it just means you have to throw $1b at RL environment companies the environments which work best in many cases have already been iterated and neatly packaged for sale, they don't depend on usage data

Samuel Hammond argues frontier AI progress is shifting from pretraining compute scale to post-training RL environments

Combined views

Samuel Hammond argues frontier AI progress is shifting from pretraining compute scale to post-training RL environments

Commentary on X

Combined views

Sentiment

Positive Read

Digg Deeper

Commentary on X

Sentiment

Positive Read

Digg Deeper