Millions of person-hours wasted building AI harnesses, erased by new model releases
sebpaquet · x · 2026-09-29
dwlz muses that hundreds of thousands—perhaps millions—of engineer person-hours were spent trying to build better AI harnesses than the labs themselves, only to be made irrelevant by each new model release. sebpaquet adds: many such efforts perform no better than simply waiting a few months.
More from AGI Musings
- Benjamin Todd's primer: three AI feedback loops could accelerate change 10-100x — ben_j_todd · 2026-09-29
- Benjamin Todd says evidence on recursive self-improvement looks 50:50 — ben_j_todd · 2026-09-29
- Data centers are more than GPU warehouses: sovereignty means the right to switch systems off — AryHHAry · 2026-09-29
- Guardian op-ed: Stop shaming AI users, organize against labor replacement — nordicinst · 2026-09-29
- AI safety drama: zero detected misalignment incidents from Chinese labs despite Western lab turmoil — basedjensen · 2026-09-29
- Mocking the ASI superpersuasion hype: catchy jingles and flashy production value — burny_tech · 2026-09-29