PhilSci Archive

The Problem of Atypicality in LLM-Powered Psychiatry

Garcia, Bosco and Chua, Eugene Y. S. and Brah, Harman (2025) The Problem of Atypicality in LLM-Powered Psychiatry. Journal of Medical Ethics.

[img] Text (https://jme.bmj.com/content/early/2025/08/07/jme-2025-110972)
The Problem of Atypicality in LLM-Powered Psychiatry (Accepted).pdf - Published Version

Download (303kB)

Abstract

Large language models (LLMs) are increasingly proposed as scalable solutions to the global mental health crisis. But their deployment in psychiatric contexts raises a distinctive ethical concern: the problem of atypicality. Because LLMs generate outputs based on population-level statistical regularities, their responses—while typically appropriate for general users—may be dangerously inappropriate when interpreted by psychiatric patients, who often exhibit atypical cognitive or interpretive patterns. We argue that standard mitigation strategies, such as prompt engineering or fine-tuning, are insufficient to resolve this structural risk. Instead, we propose Dynamic Contextual Certification (DCC): a staged, reversible, and context-sensitive framework for deploying LLMs in psychiatry, inspired by clinical translation and dynamic safety models from AI governance. DCC reframes chatbot deployment as an ongoing epistemic and ethical process that prioritizes interpretive safety over static performance benchmarks. Atypicality, we argue, cannot be eliminated – but it can, and must, be proactively managed.


Export/Citation: EndNote | BibTeX | Dublin Core | ASCII/Text Citation (Chicago) | HTML Citation | OpenURL
Social Networking:
Share |

Item Type: Published Article or Volume
Creators:
CreatorsEmailORCID
Garcia, Boscojgarca@ucsd.edu0009-0000-9279-3746
Chua, Eugene Y. S.eugene.chuays@ntu.edu.sg0000-0002-3169-7563
Brah, Harmanharman.brah@mednet.ucla.edu
Additional Information: Preprint of 8/8/2025 -- please cite published version. This article has been published in the Journal of Medical Ethics following peer review and can also be viewed on the journal’s website at 10.1136/jme-2025-110972.
Keywords: large language models, psychiatry, medical ethics, atypicality, dynamic contextual certification
Subjects: Specific Sciences > Artificial Intelligence > AI and Ethics
Specific Sciences > Medicine > Biomedical Ethics
General Issues > Ethical Issues
Specific Sciences > Medicine > Psychiatry
General Issues > Science and Society
General Issues > Science and Policy
Depositing User: Dr. Eugene Y. S. Chua
Date Deposited: 08 Aug 2025 12:47
Last Modified: 08 Aug 2025 12:47
Item ID: 25966
Journal or Publication Title: Journal of Medical Ethics
Publisher: BMJ
Official URL: https://jme.bmj.com/content/early/2025/08/07/jme-2...
DOI or Unique Handle: 10.1136/jme-2025-110972
Subjects: Specific Sciences > Artificial Intelligence > AI and Ethics
Specific Sciences > Medicine > Biomedical Ethics
General Issues > Ethical Issues
Specific Sciences > Medicine > Psychiatry
General Issues > Science and Society
General Issues > Science and Policy
Date: 7 August 2025
URI: https://philsci-archive.pitt.edu/id/eprint/25966

Monthly Views for the past 3 years

Monthly Downloads for the past 3 years

Plum Analytics

Altmetric.com

Actions (login required)

View Item View Item