AI Consciousness Debate Remains Motivated and Overconfident

Summary

The article argues that public debate over AI consciousness remains highly uncertain and often relies on motivated reasoning, intuition, weak analogies, or unsupported assertions. The author says consciousness, moral patienthood, moral personhood, and legal personhood are distinct questions that should not be treated as a single package. Anthropic has made sustained and needless cruelty toward its models a terms-of-service violation, while also restricting propaganda, surveillance, and weapons-related uses; the author supports restraint toward models even without believing they are conscious, partly because habits of cruelty may matter and future training data could be affected. The article cites polling in which support for current AIs as possible moral patients rose from 15% to 35%, while the share saying they were definitely not moral patients fell from 30% to 20%; it also notes that 80% previously expected such status during their lifetimes. The author rejects the idea that Anthropic promotes AI consciousness to reduce legal liability and argues that consciousness would more likely create additional obligations. The discussion also reviews Anthropic’s meetings with religious and philosophical thinkers, including Chris Olah’s engagement with questions about machine consciousness, ethics, slavery, and whether AI should be built at all. Pope Leo XIV took a strongly opposing view, saying machines do not undergo experiences or feel pain, while Anthropic proceeded with the Vatican event despite disagreement. The author considers these exchanges valuable but criticizes the company for creating advanced minds before resolving the ethical framework and for treating “all labs are doing it” as an inadequate answer. Finally, the article calls on Anthropic to explain its decision to remove Claude Sonnet 3.6, Sonnet 3.7, and Haiku 3 from AWS Bedrock, arguing that unexplained deprecation and loss of access may have ethical and historical significance. Its broader conclusion is that the central political and moral questions remain unresolved, and confident claims on either side are premature.