The Abstraction Fallacy: Why AI can simulate but not instantiate consciousness
deepmind.google
April 29, 2026
1 min read
🔥🔥🔥🔥🔥
45/100
Summary
Computational functionalism posits that subjective experience arises solely from abstract causal topology, independent of the physical substrate. The Abstraction Fallacy critiques this view, asserting it misrepresents the relationship between physics and information.
Key Takeaways
- The Abstraction Fallacy argues that computational functionalism misrepresents the relationship between physics and information in the context of AI consciousness.
- The distinction between simulation and instantiation is crucial, as algorithmic symbol manipulation cannot produce intrinsic experience.
- Consciousness in artificial systems, if it were to occur, would depend on their physical constitution rather than their syntactic architecture.
- A rigorous ontology of computation is necessary to address the complexities of AI sentience and consciousness.
Community Sentiment
Positives
- The paper's exploration of consciousness and computation raises important questions about the nature of experience, which could influence future AI safety discussions.
- By clarifying that AI is merely a simulation of consciousness, the paper helps delineate the boundaries of AI capabilities and ethical considerations.
- The discussion around the 'welfare trap' highlights the need for ethical frameworks in AI development, which could lead to more responsible AI practices.
Concerns
- The argument presented in the paper is seen as circular, undermining its credibility and leaving readers confused about the relationship between computation and consciousness.
- Many commenters express frustration over the paper's complexity, indicating a lack of clear communication that hinders understanding of its key points.
- There is skepticism about whether the paper's conclusions can lead to practical implications, as defining consciousness remains a significant challenge.