A korrentour readingWhat is a korrent?
Drawn from what François Chollet said
reinforcement learning Training by reward: environments, verifiable tasks, value functions, and whether it works or merely beats what came before.
Word for word, with the source under each one. They did not write this page.
Creator of the Keras deep-learning library and the ARC-AGI benchmark
Current models are unsafe not because they're too smart, but because they take goals too literally or take nonsensical shortcuts to achieve these goals, i.e. they're RL-fried. They lack common sense. They don't do the right thing in the face of ambiguity. Basically, they're not smart enough.
Where this was said
Added to korrents 13 Sept 2026 · How quotes work · Something wrong? Tell us
Our reading — they may agree, disagree or merely touch the same thing. Closest first: a shared subject counts for most, then how near the wording is.