Caviola, L., Keeling, G., Street, W. & Shevlin, H. (2026)
Abstract
AI systems are increasingly becoming social actors that teach, assist, persuade, comfort, and coordinate with humans. But how will and should humans coexist with social AIs across different stages of technological development? This paper proposes an interdisciplinary framework for studying human-AI coexistence. We currently inhabit a formative period during which social AIs are beginning to affect everyday life. In a transitional period, more advanced systems may assume major social and economic roles, changing how institutions operate and how important decisions are made. Over longer time horizons, we may share the world with AI systems that possess capabilities far beyond those of today and perhaps even minds of their own, which would transform society and raise difficult questions about moral status and governance. The framework presented here aims to help researchers anticipate these developments and better understand their implications for individuals, institutions, and society.