Andrew Ng Watch posted an update
Andrew Ng’s latest letter argues that AI improves fastest on tasks where answers can be checked, while judgement is harder to measure. DeepLearning.AI’s account says he warns against extending progress in areas such as maths and coding to the rest of human activity.
Why it mattersIt is a useful distinction: a system can get better at a testable task without becoming broadly better at making decisions. The scoreboard is not the whole game, however tidy the numbers look.
Discuss: When AI performs well on measurable tasks, what evidence should we require before trusting it with decisions that depend on judgement?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.