Tim Kellogg on Nostr: npub1fqt7n…dvxe6 on the surface, yeah we are approaching diminishing returns and so ...

npub1fqt7nf5ujj63hxjqnmgx8nw9w7te9dkxxv97mlymg0vqk06scu3smdvxe6 (npub1fqt…vxe6) on the surface, yeah we are approaching diminishing returns and so we should expect progress to slow. but that should be an indicator of AI success, but it’s framed here as failure. but also, most die-hard AI people (not me) expect progress to happen in a different model architecture. we had nearly simultaneous success with diffusion & LLMs, but future success will be something else. that in itself is rare, so we probably should adjust expectations accordingly