@haider1
Anthropic's Jack Clark says he underestimated how far AI would improve through scale alone The "bitter lesson" kept proving true: add more compute to generic neural networks, and they keep getting smarter with new abilities "specialized systems and cautious predictions keep losing to scaling"