The Bitter Lesson

An argument put forward by AI researcher Rich Sutton in 2019: methods that rely on hand-designed features and rules based on human domain knowledge will, over the long term, be surpassed by general methods that can scale with computing power.