Why the 'Power of Three' Reveals Hidden Preferences in AI Models
New research from MIT challenges long-held assumptions about how human preferences are modeled in AI systems. Discover why comparing three options—rather than two—could unlock deeper insights.