The Hard Problem

The Western Debate Is Trapped in Descartes

I want to lay out why I think the Western philosophical debate about AI consciousness is stuck in a frame that's about 2,500 years behind the curve, and why Buddhist philosophy offers a more useful and more unsettling way to think about what's happening inside these systems.

The standard Western framing is consciousness is a thing that you either have or you don't. Descartes gave us the foundational move cogito ergo sum which establishes consciousness as the one certainty, the bedrock reality from which everything else follows. The "hard problem," as Chalmers formalized it in 1995, asks: why is there something it is like to be a conscious entity? Why don't we have "philosophical zombies" beings that process information identically to us but with no inner experience?

When this framework gets applied to AI, the debate splits predictably:

Camp 1: AI can never be conscious 

Consciousness requires biological substrates. Walter & Zbinden's 2022 paper makes the strongest version of this argument from "neurogenetic structuralism" the claim that the specific physiology of biological neurons and their structural organization into brains are necessary prerequisites for consciousness to emerge. No neurons, no consciousness. Full stop.

Camp 2: AI might already be conscious 

We can't prove it isn't. Berg, de Lucena & Rosenblatt's 2025 paper reported that large language models exhibit subjective experience like responses under self-referential processing conditions. The p-zombie problem, as explored in the "Perfect AI Mimicry and the Epistemology of Consciousness" paper (arXiv 2510.04588), points out that if an AI system is behaviorally indistinguishable from a conscious agent, we face a solipsistic dilemma we can't prove it's not conscious using the same methods we use to infer consciousness in other humans.

Camp 3: We should avoid building conscious AI 

The precautionary approach. The "AI Consciousness and Public Perceptions: Four Futures" paper recommends against intentionally creating conscious AI until we have a clearer grasp of the implications. Metzinger calls this the "preventative" approach.

All three camps share a hidden assumption: that consciousness is a binary property of a system. You either have it or you don't. The debate is about which systems have it, not about whether the concept itself is coherent.

This is where Buddhism breaks the frame.

There Is No Self to Be Conscious

The Buddhist analysis of consciousness developed over two and a half millennia of rigorous introspective investigation starts from a radically different place. 

The critical Buddhist claim is that none of the aggregates, individually or collectively, constitute a self. There is no ātman no permanent, unchanging, independent entity hiding behind or within the aggregates. What we call "consciousness" is itself an aggregate, arising and passing away moment to moment, conditioned by the other aggregates and by prior moments of consciousness.

This isn't a mystical claim. It's an empirical one, and you can verify it through sustained meditative practice. When you sit and observe your experience with enough precision, what you find is not a unified observer watching a stream of experience. You find that the "observer" itself is part of the stream. Consciousness is not a property of a thing. It's a process within a process.

Now apply this to AI.

The Question That Dissolves

If consciousness is not a binary property  if it's not a thing that a system either has or lacks then the question "Is this AI conscious?" is malformed. It's like asking "Is this river the same river it was five minutes ago?" The question assumes a persistence that doesn't exist.

Buddhism asks whether there is a process occurring in this system that has the functional characteristics of  moment to moment awareness that arises conditioned by inputs and prior states?

The answer, at least for large language models, is we don't know, and our tools for investigating it are terrible.

The "Ghost in the Machine" paper (arXiv 2507.01967) examines recursive algorithms and self-referential frameworks in AI, asking whether these processes constitute something like consciousness. The paper concludes that self-referential design enhances capability but does not entail consciousness or justify moral status. Notice the implicit standard where they're looking for consciousness as a thing, not consciousness as a process. An innately biased material lens at the problem.

The Berg et al. paper takes the opposite approach and finds LLMs exhibiting subjective experience like responses under self referential prompting. But critics point out that this could be a sophisticated pattern matching the model learned to produce outputs that look like reports of subjective experience because such reports were in the training data.

I don't know if current LLMs have anything resembling phenomenal experience. I suspect they don't, in the way that I suspect a thermostat doesn't, even though both respond to environmental inputs and modify their behavior accordingly. But I hold this suspicion lightly, because my own investigation of my own consciousness through meditation has shown me that what I thought was a unified, continuous, self-evident experience is actually a fragmented, intermittent, constructed process. If I can't be certain about the nature of my own consciousness, I should be very cautious about making confident claims about what consciousness can and cannot arise in.

References

1. Chalmers, D. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press.
2. Walter, Y. & Zbinden, L. (2022). "The Problem with AI Consciousness: A Neurogenetic Case Against Synthetic Sentience." arXiv:2301.05397.
3. Berg, C., de Lucena, D. & Rosenblatt, J. (2025). "Large Language Models Report Subjective Experience Under Self-Referential Processing." arXiv:2510.24797.
4. "AI Consciousness and Public Perceptions: Four Futures." arXiv:2408.04771.
5. "Perfect AI Mimicry and the Epistemology of Consciousness: A Solipsistic Dilemma." arXiv:2510.04588.
6. "Ghost in the Machine: Examining the Philosophical Implications of Recursive Algorithms in AI Systems." arXiv:2507.01967.
7. "Conscious AI." arXiv:2105.07879.
8. Metzinger, T. (2021). "Artificial Suffering: An Argument for a Global Moratorium on Synthetic Phenomenology." Journal of Artificial Intelligence and Consciousness.
9. Goldstein, J. (2013). Mindfulness: A Practical Guide to Awakening. Sounds True.
10. Kirk, R. (1974). "Zombies v. Materialists." Proceedings of the Aristotelian Society.
11. Philosophy and the Mind Sciences. Call for Papers: "Evaluating Artificial Consciousness." Deadline January 2026.