Published November 27, 2024

NEJM AI 2024;1(12)

DOI: 10.1056/AIcs2400639

• Abstract

• Notes

• Supplementary Material

• Information & Authors

• Metrics & Citations

• Get Access

• References

• Media

• Tables

• Share

Abstract

Generative artificial intelligence (AI) models are increasingly utilized for medical applications. We tested whether such models are prone to human-like cognitive biases when offering medical recommendations. We explored the performance of OpenAI generative pretrained transformer (GPT)-4 and Google Gemini-1.0-Pro with clinical cases that involved 10 cognitive biases and system prompts that created synthetic clinician respondents. Medical recommendations from generative AI were compared with strict axioms of rationality and prior results from clinicians. We found that significant discrepancies were apparent for most biases. For example, surgery was recommended more frequently for lung cancer when framed in survival rather than mortality statistics (framing effect: 75% vs. 12%; P<0.001). Similarly, pulmonary embolism was more likely to be listed in the differential diagnoses if the opening sentence mentioned hemoptysis rather than chronic obstructive pulmonary disease (primacy effect: 100% vs. 26%; P<0.001). In addition, the same emergency department treatment was more likely to be rated as inappropriate if the patient subsequently died rather than recovered (hindsight bias: 85% vs. 0%; P<0.001). One exception was base-rate neglect that showed no bias when interpreting a positive viral screening test (correction for false positives: 94% vs. 93%; P=0.431). The extent of these biases varied minimally with the characteristics of synthetic respondents, was generally larger than observed in prior research with practicing clinicians, and differed between generative AI models. We suggest that generative AI models display human-like cognitive biases and that the magnitude of bias can be larger than observed in practicing clinicians. (Funded by Canada Research Chair in Medical Decision Sciences and others.)

View Full Text

Notes

The lead author (JW) had full access to all the data in the study, takes responsibility for the integrity of the data, and is accountable for the accuracy of the analysis. Parts of this material are based on infrastructure supported by the Ontario Ministry of Health.

This project was supported by the Canada Research Chair in Medical Decision Sciences, the Canadian Institutes of Health Research, the PSI Foundation of Ontario, and the Kimel–Schatzky Traumatic Brain Injury Research Fund. This study was supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by the Ontario Ministry of Health and the Ministry of Long-Term Care.

Disclosure forms provided by the authors are available with the full text of this article.

We thank Vidhi Bhatt, Pat Croskerry, Allan Detsky, Gordon Guyatt, Fizza Manzoor, Sheharyar Raza, Rob Tibshirani, and Robert Wachter for helpful suggestions on specific points.

Supplementary Material

Supplementary Appendix (aics2400639_appendix.pdf)

• Download
• 411.29 KB

Disclosure Forms (aics2400639_disclosures.pdf)

• Download
• 113.08 KB