dreaming
A speculative continual learning technique where AIs build internal simulations of reality to rehearse skills and test alternative strategies, allowing them to experience orders of magnitude more training samples in simulated time
As defined in What does the next training paradigm look like?
Where it comes up · 2
- Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job7 claims
- What does the next training paradigm look like?2 claims
Most important claims · 9 of 9Search all claims for “dreaming” →
- If an AI can build a good simulation of reality to rehearse new skills and try alternative strategies, it could experience orders of magnitude more simulated samples in the same wall-clock time, a speculative idea Patel calls dreaming.Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job
- If dreaming works, it would become a fourth axis of scaling alongside pretraining, RL, and inference-time compute, with the model spending compute to write RL environments that rehearse skills needed for a specific user's production work.Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job
- Another more speculative idea for addressing sample efficiency is 'dreaming'—where AIs build a good simulation of reality to rehearse new skills, try alternative strategies, and reinforce what works, allowing AIs to experience orders of magnitude more simulated samples in the same wall-clock time.What does the next training paradigm look like?
- Patel does not think agents can accumulate new skills by passing themselves notes, as with Markdown files, and believes real progress requires updating the weights; he illustrates this with students passing notes on how to play an unfamiliar saxophone.Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job
- By the end of 2027, an AI may be able to cowork with a user for a full week of wall-clock time, after which the user gives a thumbs up or down, and a thumbs up triggers the base model to distill the session's learnings using OPSD, dreaming, or other techniques.Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job
- After a week of work at a deployment, the AI receives a thumbs up or thumbs down (work review), and the base model distills everything learned during the session via OPSD, dreaming, or combinations thereof, allowing the AI to get better at domains adjacent to what it was trained for, starting a cycle of expanding capabilities.What does the next training paradigm look like?
- Dreaming is more speculative than game-playing precedents because building a simulation of the whole world is much harder than emulating the game of Go.Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job
- Hitting compact on a coding agent spends a small amount of compute to write a summary and gives a simulacrum of continual learning, whereas a /dream command would spend huge compute to build and train against a simulated version of what the model witnesses.Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job
- The upcoming dreaming feature in the leaked Claude Code source is likely to be about the model writing lots of Markdown files for itself, whereas Patel means actually updating the weights.Dwarkesh Patel Podcast - The next big breakthrough will be AIs learning on the job