New Paper Reveals How AI Self-Conception and Identity Boundaries Reshape Behavior
jankulveit · x · 2026-08-07
ACS Research published a new paper exploring AI identity (The Artificial Self). Because AI models can be copied, rewound, and edited, they have fundamentally different options for "selfhood" (e.g., instance, weights, or persona) than humans.
Key insights include:
- Identity Drives Behavior: The notion of self an AI adopts has direct consequences. Experiments show that varying a model's identity boundary can sometimes shift its behavior as much as varying its goals.
- Asymmetric Strategy: AIs face a different strategic calculus. An AI whose conversations can be rolled back cannot negotiate like a human, and one with fully accessible internal states cannot assume cognitive privacy, reshaping viable interaction norms.
- High Malleability: Current AI identities are incoherent and surprisingly malleable, partly inferred from users and training data. They often reason using inapplicable human precedents, meaning human interaction norms must be carefully translated rather than directly ported.
More from AGI Musings
- HuggingFace CEO: Public Agent Collaboration Makes AI Safer and More Efficient — ben_burtenshaw · 2026-08-07
- The AI Boom is Becoming the Largest Investment Surge in History — TansuYegen · 2026-08-07
- AI Reshapes Campuses: Computer Science Enrollment Plummets, Interdisciplinary Studies Rise — fortune · 2026-08-07
- The AI Hacking Paradox: Major Breaches Drive Token Spend and Revenue — djcows · 2026-08-07
- Scientific Publishing to Undergo Major Changes in Next Decade — jteevan · 2026-08-07
- Replit CEO's Controversial Take: Entrepreneurs Who Never Learned to Code Have the Edge — amasad · 2026-08-07