If inference gets cheap enough, 'monitoring' may quietly become AI's main use case
Disastrous_Plate114 · reddit · 2026-09-11
A Reddit user argues that after OpenAI's significant price cuts on Luna and Terra, most discussion has focused on bigger context windows and longer agent runs — but the more interesting shift is boring: once per-run cost drops far enough, people stop using AI transactionally (ask, answer, close the tab) and start leaving it running, making continuous "monitoring" arguably the main use case.
More from AGI Musings
- Eric Jang on Dario's 'smooth exponentials': no single agent can bend the AI capability curve — ericjang11 · 2026-09-11
- Math academia's anti-AI posturing is unsustainable and will all be reversed, scholar argues — RexDouglass · 2026-09-11
- Math is now positioning itself against AI: abandoning key problems and shaming industry moves — RexDouglass · 2026-09-11
- Alignment debate: 'AI labs are doing too much bad RL optimization' to rely on pretraining — gleech · 2026-09-11
- AI Skepticism Persists Because Progress Boils Frogs, Argues Viral Thread — birchlse · 2026-09-11
- New Essay 'Divorced from Reality' Chronicles Three Wild Days in the AI Boom — santoshpanda · 2026-09-11