The next big breakthrough will be AIs learning on the job

So here's the big research bet that all the labs are making. Because this kind of training will have created a kind of problem-solving agent, the kind of thing that can make progress on open-ended tasks for weeks on end in the face of errors and mistakes and ambiguity.

It goes beyond the title with direct discussion of model, learning, training, including: They think that if we train AIs to accomplish millions of verifiable tasks across thousands of diverse RL environments, then we will have basically built AGI.

1. 01Because this kind of training will have created a kind of problem-solving agent, the kind of thing that can make progress on open-ended tasks for weeks on end in the face of errors
2. 02And the people who are optimistic about this vision will say that all these things that we talk about as the fundamental deficits in the current training paradigm, for example, the
3. 03So in the previous essay, I talked about how these models are one-to-one-millionth as sample efficient as humans