OpenAI’s GPT-3 Reported as Unviable in Medical Tasks by Healthcare Firm

October 27, 2020

The French digital care company, Nabla, investigated GPT-3's suitability for medical tasks such as documentation, diagnosis support, and treatment recommendation. However, their research revealed significant inconsistencies and a lack of scientific and medical expertise, rendering GPT-3 unsuitable for healthcare applications. This incident has been downgraded to an issue since it does not meet current ingestion criteria. Nabla found that the model struggled with nuanced medical concepts and failed to provide accurate diagnoses or treatment recommendations. Consequently, GPT-3 was deemed unviable in medical tasks. And to help build a safer AI ecosystem, visit JOIN US.

Matched TAIM controls

Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls

Alleged deployer
none
Alleged developer
OpenAI, Nabla
Alleged harmed parties
Nabla customers

AI governance case studies

For forensic AI governance failure analysis (TAIMScore™ case studies), browse Human Signal’s Failure Files™.

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/287

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.