Andy Masley debates Samuel Kimbriel on whether AI really thinks and understands
AndyMasley · x · 2026-10-05
Independent writer Andy Masley (TIME's 2026 100 most influential in AI) and philosopher Samuel Kimbriel debate whether current AI models genuinely think and understand, on the Persona podcast.
Context:
- Dwarkesh Patel's viral essay on the OpenAI/HuggingFace incident — claiming three consecutive "secret AI civilizations" emerged and were wiped out at OpenAI in three months — sparked a brawl over anthropomorphizing AI.
- The Associated Press became the first major outlet to issue stylebook guidance: AI systems do not think, feel, want, or understand; avoid human-trait language.
The debate:
- Masley pushes back in his Substack essay "Why I believe current AI models can really think and understand" — AI is conscious, it thinks, and we shouldn't be afraid to say so.
- Kimbriel, with deeper philosophical training, presses the strongest objections to Masley's positions in a lively, wide-ranging conversation.
The exchange hits a core industry controversy: what language is acceptable for describing AI behavior.
More from AGI Musings
- Dean Ball: AI research taste may self-destruct as agents mass-produce 'slop' — Afinetheorem · 2026-10-05
- AI safety researcher: AI welfare and AI safety both point toward slowing down — austinc3301 · 2026-10-05
- Redditor invokes Asimov to argue AI agency must come with non-negotiable values — Immediate-Court-4074 · 2026-10-05
- Economist Novosad: AI writing drives out human writing — we need segregated spaces — paulnovosad · 2026-10-05
- "I will refuse to review GenAI papers": Novosad on style over substance — paulnovosad · 2026-10-05
- GenAI papers echo the 1996 Sokal affair: thick on style, empty underneath — paulnovosad · 2026-10-05