Philosopher Toby Ord says METR task-length extrapolation may be optimistic

Oxford philosopher Toby Ord says the 60%–85% probability range for METR's task-length doubling extrapolation seems plausible, but cautions that if larger pretrained models cause a discontinuity, the doubling pace could fall below 2x.

2026-11-10 ~ 2026-11-10 · 2 related posts

• Philosopher Toby Ord: METR's task-length doubling rate may be optimistic — tobyordoxford · 2026-11-10
• Toby Ord on METR extrapolation: pretraining jumps would mean slope change below 2x — tobyordoxford · 2026-11-10