📝 mglynnhenley wires a hallucination probe into every AI answer 5 commits 19d ago chat-uicontract-review draft
This fork scores the AI's confidence token by token, then shows users where to squint.
mglynnhenley/mikehasprobes
mglynnhenley is bolting a hallucination meter onto Mike, scoring every AI answer and showing users where to squint.
This fork takes Mike in a single, focused direction: trust calibration for AI output. mglynnhenley has wired in an external hallucination-probe service that shadow-scores assistant replies as they're generated, then surfaces that doubt visually so readers can see which parts of an answer to lean on and which to double-check.
The scoring isn't confined to chat. It runs across tabular review too, so the same confidence signal follows the user from conversational Q&A into structured document work. The presentation is deliberately quiet - a fading heat strip under the text rather than a numeric score - which reads as a product judgment that risk cues should nudge, not interrupt.
For legal-tech readers, this is the interesting question: does a continuous, glanceable hallucination signal change how practitioners actually trust an AI assistant? mglynnhenley is the one running the experiment.
What's in it
- Hallucination heat strip A fading visual band beneath each AI reply showing where the model is steadier and where it's reaching.
- Token-level confidence scoring Doubt is computed across the answer rather than as a single overall verdict, so users can localize where to squint.
- External probe service Scoring is handled by a dedicated hallucination-probe service rather than being inferred from the assistant itself.
- Coverage across chat and tabular review The same confidence signal follows users from conversational Q&A into structured row-by-row document work.
Direction
chat-uicompliance
Activity
Every assistant reply now gets shadow-scored for hallucination risk, with the doubt rendered as a fading heat strip under the text.
✅ #1 Hallucination-probe scoring for chat + tabular review +605 -16 20d ago self by
mglynnhenley ↗ analysis ↗ GitHub ## Summary
Threads of work (detailed view)
mglynnhenley wires a hallucination probe into every AI answer
This fork scores the AI's confidence token by token, then shows users where to squint.
mglynnhenley straps a hallucination meter onto Mike
Every assistant reply now gets shadow-scored for hallucination risk, with the doubt rendered as a fading heat strip under the text.
Pull requests (detailed view)
✅ Merged (1)
by
mglynnhenley · opened 20d ago · merged 20d ago by mglynnhenley