Noah Smith
· Noahpinion
· April 27, 2026 at 09:58
· ⏱ 20 min read
| Read on Substack ↗
Summary
Noah Smith argues that AI consciousness is an unanswerable question due to the problem of other minds, but proposes a research program to find neural correlates of consciousness (NCC) in humans to then replicate human-like sentience in AI. The article has no market implications—it is a philosophical essay with no trade ideas, company names, or asset-class references.
•The author introduces the 'hard problem of consciousness'—how physical processes give rise to subjective experience—and argues it can never be fully solved because of the problem of other minds.
•Smith recounts his personal experience with alexithymia (lack of subjective emotional experience) during depressive episodes, using it to argue that acting emotional and feeling emotional are distinct.
•Geoffrey Hinton believes AIs already have subjective experience, citing a thought experiment where a chatbot says 'I had the subjective experience that it was over there' after a prism bends light.
•Alexander Lerchner (Google DeepMind) argues computation is only a simulation of consciousness, not instantiation, because algorithmic symbol manipulation cannot produce phenomenal experience.
•When asked 'Are you self-aware?', ChatGPT answered no—citing functional self-reference without inner experience—while Claude answered 'I don't know,' acknowledging the hard problem applies to itself.
•Smith proposes finding the Neural Correlates of Consciousness (NCC) in humans by turning consciousness on/off with anesthesia or brain stimulation, then replicating those processes in AI to create verifiably human-like sentience.
•The author notes this research would require powerful AI to accelerate neuroscience, and that the NCC is not the same as solving the hard problem—it only identifies which brain patterns produce consciousness.
•The article concludes that debating AI consciousness is tabled for now, but will become urgent as AI becomes more autonomous, and the NCC-based approach is the best path forward despite uncertainty.