Tea@programming.dev to Technology@lemmy.worldEnglish · edit-213 days agoU.S. Citizenship and Immigration Services proposes to review social media of people applying for US citizenship, green cards, and asylum or refugee status, to comply with Trump's Executive order.www.federalregister.govexternal-linkmessage-square10fedilinkarrow-up180arrow-down13
arrow-up177arrow-down1external-linkU.S. Citizenship and Immigration Services proposes to review social media of people applying for US citizenship, green cards, and asylum or refugee status, to comply with Trump's Executive order.www.federalregister.govTea@programming.dev to Technology@lemmy.worldEnglish · edit-213 days agomessage-square10fedilink
minus-squareBzdalderon@lemmy.calinkfedilinkEnglisharrow-up6·13 days agoThis actually makes sense. The problem though, will be the same reason fact checkers failed; personal and institutional bias on what’s considered acceptable. Only an unbiased AI could do this well, and that’s a loose statement.
minus-squaredermanus@lemmy.calinkfedilinkEnglisharrow-up12·13 days agoI don’t think there’s such a thing as an unbiased AI. They’re all biased based on the input data. Something like this would be inherently subjective, it’s not a pure data question with a clear yes or no.
minus-squareCptBread@lemmy.worldlinkfedilinkEnglisharrow-up1·edit-213 days agoMachine learning and LLMs isn’t the only way to do AI though. (though your second point I do kinda agree with)
minus-squareBzdalderon@lemmy.calinkfedilinkEnglisharrow-up1arrow-down1·13 days agoI feel like you missed my “loose statement” point
This actually makes sense. The problem though, will be the same reason fact checkers failed; personal and institutional bias on what’s considered acceptable.
Only an unbiased AI could do this well, and that’s a loose statement.
I don’t think there’s such a thing as an unbiased AI. They’re all biased based on the input data.
Something like this would be inherently subjective, it’s not a pure data question with a clear yes or no.
Machine learning and LLMs isn’t the only way to do AI though. (though your second point I do kinda agree with)
I feel like you missed my “loose statement” point