Back to today’s briefing

Story perspectives

MIT Study Reveals AI Bias Threatens Patient Care Equity

8/29/2025

47 4

Summary

Story summary

MIT research by Marzyeh Ghassemi indicates AI tools in hospitals are vulnerable to misleading inputs, jeopardizing patient care.
AI models demonstrate reduced accuracy when faced with modified medical questions.
Ghassemi calls for diverse training data and regular audits to address biases against women and minorities in AI diagnostics.
She stresses the importance of regulation to promote equity in clinical AI applications.

1 of 1

Story summary

MIT research by Marzyeh Ghassemi indicates AI tools in hospitals are vulnerable to misleading inputs, jeopardizing patient care.
AI models demonstrate reduced accuracy when faced with modified medical questions.
Ghassemi calls for diverse training data and regular audits to address biases against women and minorities in AI diagnostics.
She stresses the importance of regulation to promote equity in clinical AI applications.