Toward Criteria for Artificial Self-Consciousness: Unity, Normativity, and Agency
Proceedings of the AAAI Symposium Series May 18, 2026 Peer reviewed DOI: 10.1609/aaaiss.v8i1.42563 via OpenAlex
Summary
The paper distinguishes two types of consciousness often confused in AI debates: pre-reflective experiential awareness (the minimal self-involvement in any experience) and reflective self-consciousness (a unified standpoint enabling commitments, evaluation under norms of truth and value, and revision in light of reasons). Reflective self-consciousness is analyzed through agency, normativity, and unity, with a structure of epistemic answerability involving commitment formation, persistence, conflict detection, and error revision. This distinction clarifies ethical issues around artificial consciousness and suggests that emerging AI systems may challenge traditional moral categories for moral standing.
Study at a glance
| Design | theoretical or philosophical paper |
|---|---|
| Key finding | Pre-reflective experiential awareness and reflective self-consciousness are distinct forms of self-involvement, and their conflation obscures how emerging artificial systems may pressure inherited moral categories for moral standing. |
Abstract
This paper distinguishes two forms of consciousness that are often conflated in debates about artificial intelligence: pre-reflective experiential awareness and reflective self-consciousness. Drawing on phenomenology, Kant, and con-temporary philosophy of mind, it argues that pre-reflective awareness involves the minimal self-involvement characteristic of phenomenal experience, while reflective self-consciousness involves a unified standpoint from which a subject can form commitments about how things are, evaluate them under norms of truth and value, and revise them in light of reasons. The paper analyzes reflective self-consciousness in terms of agency, normativity, and unity, articulating a structure of epistemic answerability that includes commitment formation, persistence across time, conflict detection, and re-vision in response to error. Distinguishing these two forms of self-involvement clarifies the ethical landscape of artificial consciousness and suggests that emerging artificial systems may pressure the inherited moral categories through which moral standing has traditionally been understood.