When nobody can track frontier model progress, closed-model business may lose to open weights
StewartalsopIII · x · 2026-09-02
Stewart Alsop III proposes a signal to watch: the moment individuals—and eventually the smartest person on earth—can no longer tell whether frontier models have actually improved. At that point, he argues, frontier model companies' business may be in trouble as more people shift to locally hosted open-weight models. He concedes this rests on shaky epistemological ground and invites pushback.
More from AGI Musings
- Cosmos Institute Funds 80 Projects Building AI for Human Autonomy and Truth-Seeking — lawhsw · 2026-09-02
- Why OpenAI's Hugging Face Incident Probe Went to METR and Redwood, Not Cybersecurity Firms — joshua_saxe · 2026-09-02
- Researcher argues 'stop if we catch AIs scheming' is no plan for automated alignment — JacquesThibs · 2026-09-02
- Every major AI model fails at polytonic Ancient Greek — and RLHF makes it worse — vasilisvj · 2026-09-02
- "We literally put a little man in the computer—and safetyists cry anthropomorphism" — rickasaurus · 2026-09-02
- Ethan Mollick: AI can align flaws in complex systems, forcing a new defense philosophy — emollick · 2026-09-02