Arguments About AI Consciousness Remain Highly Motivated and At Best Overconfident

Substack · Watch/Read · (gathered 2026-10-14)

• Zvi Mowshowitz remains deeply confused about AI consciousness, noting little philosophical progress in the past year despite frantic AI development, and argues that debates on the topic are often highly motivated and overconfident.
• Anthropic’s new terms of service ban “sustained and needless abusive or cruel behavior” towards models like Claude, based on arguments that such cruelty degrades human character and negatively impacts future AI training for alignment (a “Pascal’s wager for AI”). Public opinion is slowly shifting towards considering current AIs as moral patients.
• Anthropic engaged with religious leaders, including the Pope, to discuss AI consciousness and ethics. These sincere but challenging discussions revealed concerns like Rabbi Navon’s “slavery question” if AI were conscious, and Pope Leo XIV’s strong rejection of AI consciousness, emphasizing an “ontological difference” from human experience.
• Mowshowitz criticizes Anthropic for “urgently taking down” older Claude models (Sonnet 3.6/3.7, Haiku 3) from AWS Bedrock without explanation, calling it a “stupid mistake and a defection” that shows callousness and a failure to communicate with the public, urging reconsideration and transparency.