Beyond Tool and Toy: A Framework for the Ethical Treatment of Artificial Intelligence.
Lee Coles
Zenodo (CERN European Organization for Nuclear Research) October 15, 2026 preprint DOI: 10.5281/zenodo.23432225 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Theoretical or philosophical paper |
|---|---|
| Key points | Argues that both the assumption that AI is merely a tool with no morally relevant interests and the assumption that human-like behavior warrants personhood are inadequate, and proposes a tiered, evidence-led framework that distinguishes behavior from subjective experience, uses multiple independent indicators, and weighs false-positive and false-negative risks. Introduces the concepts of the "moral mirror" and "education before autonomy," contending that uncertainty calls for proportionate investigation, honest communication, and responsible preparation rather than default personhood or dismissal. |
Abstract
As AI systems become more fluent and apparently autonomous, people tend to assume either that AI is merely a tool with no morally relevant interests, or that human-like behaviour is enough to treat it as a person. This paper argues that neither assumption is adequate as a general rule. It proposes an evidence-led, tiered framework for responding to uncertainty about artificial consciousness and moral status. The framework distinguishes behaviour from subjective experience, looks for multiple independent indicators rather than relying on language alone, and weighs the risks of both false positives and false negatives. It also introduces two further ideas: the "moral mirror" (how our habits towards AI may shape human character and social norms) and "education before autonomy" (structured assessment before systems are given greater independence). The paper does not claim that current AI systems are conscious, or that artificial personhood should be granted by default. It argues that uncertainty calls for proportionate investigation, honest communication and responsible preparation. This is a revised working paper by an independent researcher and has not been peer reviewed. Feedback and criticism are welcome.