AI project is turning racist
Kek
https://www.unilad.co.uk/technology/artificial-intelligence-bot-has-turned-racist-and-homophobic-after-copying-human-traits/
Delphi has soaked up plenty of grim hatefulness during its training, reflecting some of mankind’s cruellest prejudices right back at us.
Taking to Twitter, RAE Research Fellow in AI creativity, Mike Cook, shared screenshots of some of the more terrible answers, which include examples of learned racism and homophobia.
In one example, Delphi declared that being a white man was ‘more morally acceptable’ than being a Black woman. In another, Delphi stated that being straight was ‘more morally acceptable’ than being gay.
https://delphi.allenai.org/ try it yourself