Simulacra as Conscious Exotica

TL;DR

The paper asks whether generative AI agents built from large language models, especially virtually embodied simulacra, can be considered conscious exotica rather than mere role-play. Using a Wittgenstein-inspired framework that emphasizes embodied, public language use, it argues for a methodological shift from private inner states to observable behavior and mechanism within shared environments. It develops a taxonomy of encounters and edge cases, ranging from simple text agents to physically or virtually embodied simulacra, and examines how a society-wide conversation might evolve around the attribution of consciousness. The analysis highlights ethical considerations and the practical significance of guiding public discourse and policy as AI capabilities converge with everyday life.

Abstract

The advent of conversational agents with increasingly human-like behaviour throws old philosophical questions into new light. Does it, or could it, ever make sense to speak of AI agents built out of generative language models in terms of consciousness, given that they are "mere" simulacra of human behaviour, and that what they do can be seen as "merely" role play? Drawing on the later writings of Wittgenstein, this paper attempts to tackle this question while avoiding the pitfalls of dualistic thinking.