references
Curated essays and papers that scaffold the research direction. Each card links to the source; longer reading-archive write-ups live under /blog/category/papers/.
-
The Bitter Lesson
Seventy years of AI research show that general methods leveraging computation are, in the long run, the most effective — and by a large margin.
The single most-cited modern essay on research direction in AI. The core observation — that scaling generic computation eventually beats encoding clever human priors — is the empirical scaffold for almost everything downstream: large language models, RL scaling laws, the agent-loop work, and the Discovery + Selection axes of the framework in the Enhanced POET reading-archive entry. Worth re-reading periodically to recalibrate.