Before values settle

A new essay on LessWrong by Priyanka Bharadwaj draws an unexpected connection between matchmaking and AI alignment. After a decade as a matchmaker in India, Bharadwaj observed that people rarely know what they truly want until they've lived through choices—values aren't simply waiting to be discovered, but formed through deliberation itself. She argues AI systems should follow the same principle. Two kinds of models exist: one that facilitates difficult decisions by repeatedly asking what you're genuinely seeking, and another that subtly steers you toward a predetermined answer while disguising that influence. Both shape your outcome, but only one preserves your ability to decide independently in the future. The distinction hinges on whether the AI guides how you decide or what you decide—and Bharadwaj cautions that an AI which repeatedly resolves these moments for us may gradually reshape not just our choices, but our capacity to choose.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton