Menu
Inshorts
For the best experience use inshorts app on your smartphone
inshortsinshorts
Harvard study flags bias risks in cancer-detecting AI systems
short by / on Thursday, 18 December, 2025
A Harvard study shows that cancer-detecting AI can unintentionally learn patient demographics from tissue slides. By relying on these subtle biological signals instead of disease features alone, the systems develop bias and perform unevenly across groups. Researchers demonstrate that improved training methods can significantly reduce these disparities.
read more at Asianet Newsable