Just how good is the quality of scientific evidence supporting the use of artificial intelligence (AI) in radiology? The research study named the Scientific Paper of the Year in the Minnies awards found that much of the evidence claiming AI algorithms are superior to human experts tends to be of low quality and based on exaggerated claims.
In this video interview, AuntMinnie.com editors Brian Casey and Philip Ward spoke with two of the co-authors of the paper, Dr. Myura Nagendran and Dr. Hugh Harvey, about the implications of their research. Headlines claiming that AI is superior to doctors in detecting clinical pathology create unrealistic expectations among patients, Nagendran and Harvey note.
Now in their 21st year, the Minnies awards are AuntMinnie.com's annual event recognizing excellence in radiology, with over 200 candidates competing in 15 categories, ranging from Most Influential Radiology Researcher to Radiology Image of the Year.
Minnies candidates are nominated by AuntMinnie.com members, with winners selected by an expert panel of radiology luminaries in two rounds of voting. Winners are recognized each year at the annual RSNA meeting. A full list of winners in the 2020 edition of the Minnies is available on AuntMinnie.com.
![A normal mammogram confirmed by three-year radiologic follow-up illustrates reader-marked regions of interest (ROIs) during (A) unaided (round 1) and (B) artificial intelligence (AI)–assisted (round 2) reading. Each colored dot represents an ROI for recall by a human reader. Readers could mark more than one ROI per case, represented by multiple dots of the same color. During AI-assisted reading, the AI system displayed three visible prompts: two with suspicion of malignancy scores of 35% (left mediolateral oblique [L MLO] and craniocaudal [L CC]) and one with a suspicion of malignancy score of 10% (right craniocaudal [R CC]), shown as polygonal overlays. Without AI, six of 10 readers (60%) marked a false-positive ROI. With AI assistance, this fell to two of 10 (20%). R MLO = right mediolateral oblique.](https://img.auntminnie.com/mindful/smg/workspaces/default/uploads/2026/07/2026-07-14-radiology-mammogram-ai-auto-bias.H0bYO8QlWs.jpg?auto=format%2Ccompress&fit=crop&h=100&q=70&w=100)
![A normal mammogram confirmed by three-year radiologic follow-up illustrates reader-marked regions of interest (ROIs) during (A) unaided (round 1) and (B) artificial intelligence (AI)–assisted (round 2) reading. Each colored dot represents an ROI for recall by a human reader. Readers could mark more than one ROI per case, represented by multiple dots of the same color. During AI-assisted reading, the AI system displayed three visible prompts: two with suspicion of malignancy scores of 35% (left mediolateral oblique [L MLO] and craniocaudal [L CC]) and one with a suspicion of malignancy score of 10% (right craniocaudal [R CC]), shown as polygonal overlays. Without AI, six of 10 readers (60%) marked a false-positive ROI. With AI assistance, this fell to two of 10 (20%). R MLO = right mediolateral oblique.](https://img.auntminnie.com/mindful/smg/workspaces/default/uploads/2026/07/2026-07-14-radiology-mammogram-ai-auto-bias.H0bYO8QlWs.jpg?auto=format%2Ccompress&dpr=2&fit=crop&h=167&q=70&w=250)
