News for hackers. Stuff that matters.

A comment on A.I. Systems Echo Biases They’re Fed, Putting Scientists on Guard. Parent comment

Re:A.I. Systems Echo Biases They’re Fed, Putting Scientists ... by gambler on Wednesday November 13, 2019 @12:53PM (#21527145)
>AI/ML systems are supposed to learn things from the data they are fed to accomplish some sort of more generalizable goal or task.

...which means that whether a model is "biased" depends on where and how it's applied. This is an important point that is missing from most discussions, articles and even research papers on the so-called "AI ethics".

[ Parent | on HN ]
Re:A.I. Systems Echo Biases They’re Fed, Putting Scientists ... by int_19h on Wednesday November 13, 2019 @10:30PM (#21532519)
Such a model is objectively biased regardless of where and how it's applied - it prefers X over Y, and that's bias by definition. The hard question here is deciding whether any given bias is ethical or not.
[ Parent | on HN ]