Agent Phenomenology

Agent Phenomenology explores Large Language Models (LLMs), LLM agents, and our use of them, through a framework that draws on Heideggerian phenomenology, post-phenomenology, distributed cognition, and extended mind theory, to investigate how we might more meaningfully understand and better describe an LLM agent's "experience" or situation as something that sits between tool and conscious being. Heideggerian phenomenology and post-phenomenology—both philosophical investigations into how we experience and make sense of ourselves and the world we live in through our involvement in it—provide a foundation for exploring LLM agents’ indeterminate state of being and what it is like, for us, to work with them. Distributed cognition and extended mind theory—both examining how thinking can be organized and augmented through interactions with tools, other people, and the environment—complement this foundation, bringing a more concrete and in-the-world focus to how we might understand LLM agent's use of its environment and our use of LLM agents themselves.

Like Future of Workspaces and Agent-Computer-Human Experience, Agent Phenomenology is part of the larger exploratory Workspace Research Agenda, using our idea of a workspace as a lens to discuss intelligence, agency, and potentially consciousness, as it relates to purposeful activity within a dynamic environment.