AI Tools for Breast Cancer Detection Fall Short of Radiologists' Expectations

A survey published in Clinical Imaging polled 215 members of the Society of Breast Imaging about their experience with AI tools for breast cancer detection. The results show that the tools work but less than hoped: 35% of users report decreased recall rates (expected: 59%), 9% report fewer unnecessary biopsies (expected: 36%), and 29% report less burnout (expected: 56%). Most radiologists view AI primarily as a second opinion. Costs and lack of institutional support remain the biggest hurdles. The research team led by first author Joud Almogati of UC San Diego Health noted that only a few consider AI's contribution decisive. The high expectations may be partly self-inflicted by the AI industry, as prominent AI researchers predicted a decade ago that radiologists would soon lose their jobs. Today, similar statements are heard about knowledge work on computers, with Jensen Huang speaking of a God complex among AI job-loss prophets.