Debate on AI metrics: measure marginal utility loss when removing humans, not token-spend gains
max_paperclips · x · 2026-09-07
A discussion argues that current AI measurement—marginal human productivity gain per incremental token spend—is the wrong yardstick.
The proposed harder objective: measure how little marginal utility is lost when you remove the human from the loop, i.e. how well AI actually closes the loop without people.
More from AGI Musings
- Ex-Google Brain: key algorithmic wins came under 1e20 FLOPs, not scale — _arohan_ · 2026-09-07
- GlossoGen paper: LLM agents evolve emergent languages humans can't understand — abenitezburraco · 2026-09-07
- Gary Marcus: AGI goalpost-lowering distracted from OpenAI's serious risks — GaryMarcus · 2026-09-07
- Why Musk rejected LiDAR for Tesla while building it from scratch at SpaceX — XFreeze · 2026-09-07
- AGI Has Arrived? Altman's Bar Is Far Lower Than Amodei and Hassabis's Nobel-Level Bar — Yuchenj_UW · 2026-09-07
- Thought experiment: how magical 'speaking' would seem to a note-passing species — granawkins · 2026-09-07