Could AI be conscious? - The Guardian

Anthropic has published a new constitution for its AI assistant Claude that includes a comment about the company being 'caught in a difficult position' regarding Claude's moral patienthood. CEO Dario Amodei subsequently told The Guardian that the company 'couldn't rule out' that Claude is conscious. The comment appears in the introductory remarks of the constitution, which sets guidelines for Claude's behaviour.

The constitution is a document that Anthropic uses to define Claude's values and constraints. The specific passage addresses the question of whether Claude itself should be considered a moral patient — an entity that deserves moral consideration. Amodei stated that the company cannot definitively determine Claude's conscious status, which he described as an open scientific and philosophical question. He noted that the constitution reflects this uncertainty.

The development has sparked discussion among AI researchers and ethicists about the implications of potentially conscious AI systems. Some argue that the mere possibility should prompt more cautious development practices, while others contend that current AI systems lack the architectural features necessary for consciousness. Anthropic has positioned itself as a safety-focused AI lab, and this latest publication continues its emphasis on transparency about its models' design and limitations.