Concerns over AI surveillance in schools are intensifying after armed officers swarmed a 16-year-old student outside Kenwood High School in Baltimore when an AI gun detection system falsely flagged a Doritos bag as a firearm.
Allen was handcuffed at gunpoint. Police later showed him the AI-captured image that triggered the alert. The crumpled Doritos bag in his pocket had been mistaken for a gun.
Well… first we need to know how that shit was trained. I’d guess, either they take rl pics and let humans decide arbitrarily, they take historic data of surveillance pics + if bastards were sent out, or they use historic pics + if something was actually found/confirmed.
All three will be heavily biased, especially #1. #2 would at least be based on “experts’” decisions, and #3 would be the least biased, but still way too much to be a basis for anything.
Because: Human behaviour can be fixed. With the right measurements, biased people/racists can be retrained or taken off-duty. However, as soon as such a system is trained on biased data, gl correcting that. And no one will feel themselves responsible anyway.
This is the scary thing about the development of any kind of AI we are building as a society generally … it will be heavily biased according to our general ideas of race, identity and beliefs … no matter how you cut it, develop it, influence it or deal with it, the AI will always take away our human based general ideas of right, wrong, biases, beliefs, and perceptions.
When you think about it, we (humanity as a whole) are trailer trash morons who dropped out of school years ago that shouldn’t have the responsibility of raising a child … but we went out and got pregnant anyway and now we’re raising a baby … sure, there are instances of trailer trash parents who do go out to raise decent people but I’ve seen my share of down and out people who raise children in absolutely the worst ways possible and churn out a whole generation of maladjusted people who end up with drug addictions, and their only hope is in doing the absolutely dumbest things possible to get by in life.
We’re dumb parents and we’re raising a new being-AI-intelligence-whatever into existence. What do you think that new child (or growing AI) will become?
It’s shocking just how biased the third is despite sounding (relatively) reasonable and unbiased. But bias is the entire nature of AI, that’s the whole point: biasing a machine to transform and distill data to arrive at a desired output.
Really, the instant anything is made into a process with or without AI there’s bias, and it’s inescapable.
ETA: I guess what I’m getting at is
is bias in and of itself