Google's Haydn Belfield says AI is not yet self-improving, citing DeepMind data
Overview
Google engineer Haydn Belfield says DeepMind sees no measurable acceleration of AI research driven by AI itself.
He points to figures showing AI writes more than 75 percent of production code, while about 88 percent of changes still need human review, and that success without human input falls as task complexity rises. A General Capability Index covering 123 models shows steady growth of about 0.1 points per month with no significant speedup. Belfield's view contrasts with warnings from more than 20 researchers, including Geoffrey Hinton and Yoshua Bengio, of a possible intelligence explosion.
Written by AI from the articles below · updated Oct 14, 3:49 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- The DecoderNewsGoogle engineer says self-improving AI isn't close yet
AIGoogle engineer Haydn Belfield says DeepMind sees no measurable acceleration of AI research driven by AI itself. AI writes more than 75 percent of production code, but about 88 percent of changes still need human review, and success without human input falls as task complexity rises. A General Capability Index covering 123 models shows steady growth of about 0.1 points per month with no significant speedup, while more than 20 researchers, including Geoffrey Hinton and Yoshua Bengio, have warned of a possible intelligence explosion.
Heat trend
Not enough continuous observations to show a trend yet.