When silence is safer: a review and decision-theoretic framework for LLM abstention in healthcare
Oriana Presacan, Alireza Nik, Jaya Ojha, Vajira Thambawita, Bogdan Ionescu, Michael A. Riegler · 2026
Read When silence is safer: a review and decision-theoretic framework for LLM abstention in healthcare by Oriana Presacan, Alireza Nik, Jaya Ojha, Vajira Thambawita, Bogdan Ionescu, Michael A. Riegler (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13631354.txt.
Description
Oriana Presacan and five colleagues review when medical chatbots should stay silent. Their review appeared in 2026. Language models are built to answer. In healthcare a confident wrong answer can injure. So the authors survey research on abstention. That means refusing or deferring when the model is unsure. It also covers answers that could themselves cause harm.
They find most current safeguards bolt on outside tools. Few teach the model restraint itself. Top models still fail to refuse unfit medical prompts. Almost no benchmarks test abstention in realistic clinical conversations. To move the field, they formalise the answer-versus-withhold choice in decision theory. They also pilot MedSAFE, an evaluation framework built from clinical scenarios drawn from their review.
Catalog record
| CREATOR | Oriana Presacan, Alireza Nik, Jaya Ojha, Vajira Thambawita, Bogdan Ionescu, Michael A. Riegler |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 113281 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |