Bitter Lesson of AI

See an interesting post by Richard Sutton on his blog incomplete ideas. The article describes progress in AI which advances by enabling models to learn rather than teaching them what they should "know". A good point in the light of the current AI "arms race" where many different approaches are being pushed out to market for human use, possibly too soon for some models.

Previous
Previous

Writing Code with AI Help

Next
Next

Do Robots need Passports?