khurram javed

The 'big world hypothesis' posits that synthetic data generation is bottlenecked by human expertise; true intelligence requires agents that learn models from their own experience and plan with self-discovered abstractions, as animals do, rather than relying on human-engineered simulations.

The 'big world hypothesis' posits the world is infinitely complex; synthetic data and simulations are bottlenecked by human expertise. True intelligence requires agents that learn their own world models from raw experience, discover abstractions, and plan with them — eliminating the human-in-the-loop for data curation and simulator maintenance.

Current systems lack the ability to learn world models from experience and then plan with those self-discovered abstractions. This model-learning + planning loop is the key to paradigm-shifting intelligence and is the core focus of Oak Lab's Alberta Plan.

Oak Lab targets a trillion-parameter continually learning model running at 20 watts within 5-10 years, relying on two orders of magnitude compute efficiency improvement (Moore's Law) plus algorithmic breakthroughs. Current labs are locked into energy-intensive scaling; a paradigm shift to efficient continual learning could unlock massive energy savings.

Oak Lab aims for a trillion-parameter model consuming 20 watts by leveraging two orders of magnitude compute efficiency gains from Moore's Law over 5-10 years combined with algorithmic breakthroughs in continual learning, challenging current energy-intensive scaling paradigms.