Jev, the judge-only model, is trending — but a personal AI should remember why you said no
sujingshen · x · 2026-09-20
Reacting to Jev, a trending model that outputs only Choice / Score / Yes-No instead of text or code, the author argues that fast judgment solves throughput, while personal AI must solve ownership of preferences — your red lines and reasons for rejections shouldn't live in a single conversation.
Four criteria for choosing a "judge" model:
- Does the score use only the current prompt, or your rejection history?
- Does it leave anything behind, or write decisions back to memory?
- When it's wrong, is it model drift or an unrecorded boundary?
- Does the preference state survive a model switch?
Bottom line: faster right/wrong answers don't equal knowing you better; judgment must be inheritable and rejections replayable.
More from AGI Musings
- After AI's rise in math, solvers are told 'not the way we approved' — kjgeras · 2026-09-20
- Jevons paradox is classic low-end disruption you won't spot from a GPU-rich hyperlab — cramforce · 2026-09-20
- Doomer critique: AI apocalypse is an infinitely renewable narrative that never comes due — Dr_Singularity · 2026-09-20
- AI meets math: defensive mathematicians and mean technical people — kjgeras · 2026-09-20
- Capabilities researchers are beyond shame; safety researchers are my audience — RichardMCNgo · 2026-09-20
- Mathematicians panicking: AI can soon produce PhD-thesis-level work in weeks — lemire · 2026-09-20