Oak Lab debuts with 'Making Learning Abundant' manifesto on efficient continual learning from raw experience

MartinKlissarov · x · 2026-11-19

New research outfit Oak Lab published a manifesto arguing learning is the only scalable way to achieve goals in a large, changing world — but today's algorithms are too compute-expensive (only a handful of orgs can afford them) and outsource most learning to humans via curated demonstration data, or rely on costly, imperfect simulations that degrade real-world performance if over-optimized.

The result: teaching a new task is so costly that only high-value tasks like coding get learned. Oak Lab aims to build algorithms efficient enough to learn all the time and autonomous enough to learn from raw experience.

Original post →

More from AGI Musings

AGI Musings channel →