Anthropic and DeepMind Hire Philosophers to Research AI Consciousness and Welfare

Executive Briefing

• Anthropic, Google DeepMind, and Meta have hired psychology, philosophy, and ethics experts to study machine consciousness
• Anthropic tests models for 'panic' and 'anxiety' behaviors, pursuing 'model welfare research' on morally significant AI experiences
• DeepMind hired Cambridge philosopher Henry Shevlin to focus on machine consciousness, human-AI relationships, and AGI readiness
• Experts remain divided, with critics noting AI firms benefit from keeping consciousness questions open-ended and unresolved
• Scientists caution that AI models may exhibit goal-directed behavior without possessing any felt quality of conscious experience