Login
You're viewing the mstdn.social public feed.
  • Unnamed TNG skant beefcakeresearchfairy@scholar.social
    Nov 24, 2025, 6:32 PM

    The problem with crime prediction algorithms is that they're racist false-positive machines that just make life worse for black people; not "Is it unjust to arrest someone for a crime they didn't commit, but would?"

    The problem with AI is stuff like: colonial exploitation, theft, environmental damage; not that robots become so good that they gain personhood or overthrow us

    The problem with eugenics is devaluing of humans now; not a hypothetical perfect superhuman

    💬 3🔄 12⭐ 27

Replies

  • OggieOggie@woof.group
    Nov 24, 2025, 6:35 PM

    @researchfairy

    There's an interesting little subset story of an Asimov thing that got almost no time in the books, but is interesting, because it -broke- the three laws, completely

    A robot was constructed to think only humans who spoke with a specific accent were human. Everyone else was perfectly fine to kill.

    I think about that meduim often, because it really does approach 'everything wrong with the current real world scope' a lot better.

    💬 0🔄 1⭐ 0
  • Nege :GOOSE_PNG:theynege@beige.party
    Nov 24, 2025, 6:45 PM

    @researchfairy people think that data is value neutral, objective and unbiased and it's only the human interpretation that skews it. this is like the notion I have unfortunately seen more than once that society is rebuilt for every child born, and thus there is no accumulated mess of systemic oppression for them to face; it's all personal choice and responsibility for using the equal options given to all human beings. it's humiliating to still see people think that capitalism is freedom if you have the gumption and good character, and so what LLMs are exposed to in the production of their slop has merit in truth because of how it was achieved or created.

    there's just no clear place to start in dissecting this.

    💬 0🔄 0⭐ 1
  • 💬 0🔄 0⭐ 0