News for hackers. Stuff that matters.

A comment on Toward Guarantees for Clinical Reasoning in Vision Language Models.

Re:Toward Guarantees for Clinical Reasoning in Vision Langua... by barthelomew on Sunday March 01, 2026 @09:57PM (#47213340)
AI (VLM-based) radiology models can sound confident and still be wrong ; hallucinating diagnoses that their own findings don't support. This is a silent, and dangerous failure mode.

Our new paper introduces a verification layer that checks every diagnostic claim an AI makes before it reaches a clinician. When our system says a diagnosis is supported, it's been mathematically proven - not just guessed. Every model we tested improved significantly after verification, with our best result hitting 99% soundness.

We're excited about what comes next in building verifiably correct AI systems.

[ Parent | on HN ]
Re:Toward Guarantees for Clinical Reasoning in Vision Langua... by phoenixrecruit on Sunday March 01, 2026 @10:05PM (#47213378)
nice work!I know about your work, similar to this: https://arxiv.org/abs/2601.20055 and https://github.com/DebarghaG/proofofthought
[ Parent | on HN ]
Re:Toward Guarantees for Clinical Reasoning in Vision Langua... by barthelomew on Sunday March 01, 2026 @10:24PM (#47213489)
Yes, indeed! This work uses the Proof of Thought library and several techniques from VERGE!
[ Parent | on HN ]