@ylecun@PSpagnou I really like this analogy. Chess gadget is a one trick pony, but LLMs are N-trick ponies. N grows with time/research/scale. When N gets large it doesn’t matter if it’s really human-level in all the aspects. It will beat X% of people in Y% of tasks.
@PSpagnou The $30 gadget that can beat you at chess "outperforms human-level performance in some task", yet is clearly not on the path towards Human-Level AI.
"Soon, you’ll see AI-powered features in [Google] Search that distill complex information and multiple perspectives into easy-to-digest formats" is the most underrated sentence I've ever read.
overheard at Again Bar: "Decentralisation brings its own major issues, it’s not a silver bullet. Thinking that Mastodon-like services will take over is a pipe dream."