Dwarkesh's Bet: Sample Efficiency and Continual Learning Will Step-Change AI Within 1-2 Years

PTrubey · x · 2026-09-30

PTrubey endorses Dwarkesh's read of the next 1-2 years: while OpenAI and Anthropic squeeze maximum efficiency out of the existing backpropagation paradigm, others are tackling core ML problems like sample efficiency and continual learning — breakthroughs that could step-change the entire industry. The point draws on Dwarkesh's recent interview with an OpenAI researcher, highlighting a split between in-paradigm optimization and paradigm-level research.

Related event: OpenAI Researcher: Sample Efficiency and Continual Learning May Upend Backprop Paradigm(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →