Researchers from UC Berkeley have discovered racial prejudice has somehow been programmed into an algorithm used nationwide in the USA for deciding who would benefit from medical interventions. White patients were judged as equally ill as black patients who were in fact sicker – meaning treatment was awarded to white patients far more often. Not a racist AI entity, but one which was coming to a prejudiced conclusion, based on data fed to it. The problem has 80% been fixed, but underlines the importance of not over-relying on the infallible logic of computers. Read more here.
Today’s Simple AI – AI made easy