Richard S. Sutton is a computer scientist widely regarded as one of the founders of modern reinforcement learning, the branch of machine learning in which an agent learns by trial and error from rewards. He co-authored the standard textbook on the subject and developed core ideas such as temporal-difference learning that underpin systems from game-playing agents to robotics.
Sutton is also known beyond his technical work for a short, widely read essay, “The Bitter Lesson,” published on his website incompleteideas.net on March 13, 2019. Its central claim is that “the biggest lesson that can be read from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin.” He argues that approaches relying on built-in human knowledge tend to be overtaken, over time, by simpler general methods such as search and learning that scale with ever-cheaper computing power.
The Bitter Lesson became a touchstone for the era of large models trained on massive compute. It is frequently cited to explain why scaling up general learning systems, rather than hand-engineering knowledge, has driven so many recent breakthroughs.