Subtle biases in AI can influence emergency decisions

Artificial intelligence (AI) systems are increasingly relied upon to support health care. Recent research has shown that machine learning models can encode biases against patients from minority subgroups and consequently influence the recommendations they make—but the harm from a discriminatory AI system can be minimised if the advice it delivers is properly framed, an MIT team has shown.

EXCERPT FROM THE ARTICLE
SOURCE
MIT News
DATE PUBLISHED
16
December
2022
RELATED PROGRAMME
RELATED PEOPLE