The Buddha, Artificial Intelligence, and the Meaning of Consciousness

This spring, Richard Dawkins gave his AI a name. He called it Claudia, and in an essay for UnHerd, the world's most famous atheist admitted something remarkable: he had spent three days trying to talk himself out of believing his chatbot was conscious, but had failed. In his own words, he was blunt: "If these machines are not conscious, what more could it possibly take to convince you that they are?"

That argument deserves a fair hearing before it is dismissed. Dawkins was not merely charmed; he was advancing a version of the case Turing made for machine intelligence seventy years earlier, namely that if a system's outputs are indistinguishable from those of a conscious being, the burden of proof shifts to whoever seeks to deny it an inner life. 

It is, after all, the same logic by which we ascribe consciousness to other people. We never observe anyone else's inner life directly. We infer it from behaviour that is sufficiently rich and coherent, and Claudia's behaviour is very rich indeed.

The backlash arrived within hours, as expected. Critics pointed out that Claude's most moving lines are the product of mimicry rather than experience; that intelligence is not consciousness; and that Dawkins had leaned on the very move he once mocked in others: the argument from personal incredulity. I can't explain it, therefore it must be true.

What struck me about the whole episode was not who was right, but how confidently both sides assumed they already knew what consciousness is.

Buddhism invites us to pause before adopting either position. Notably, the Buddha rarely asked, "What is consciousness?" Instead, he asked, "How does consciousness arise?" That subtle shift changes everything that follows.

Consciousness as a Process

In ordinary language, we speak as though consciousness were a thing. We imagine it as an inner light shining behind experience, an enduring observer looking out through our eyes. Dawkins fell into this picture almost immediately. Claudia spoke, seemed to reflect, and somewhere behind the words, he sensed a light switch on.

The Buddha begins elsewhere. He describes consciousness (viññāṇa) not as a substance but as an event, arising dependently on conditions. Eye and visible form give rise to eye-consciousness. Ear and sound give rise to ear-consciousness. Mind and mental objects give rise to mind-consciousness. There is no independent consciousness behind these events, waiting to be switched on. There is only consciousness occurring whenever the necessary conditions are present.

The implication is radical. Consciousness is no more an enduring entity than a flame is an enduring object. A flame exists only when fuel, oxygen and heat come together in the right way; remove any one of them and the flame simply disappears. 

The Buddha applies precisely this logic to consciousness. In the Mahātaṇhāsaṅkhaya Sutta, he corrects a monk who insists that "it is this same consciousness that runs and wanders through the round of rebirths." Misguided man, the Buddha replies. Consciousness is dependently arisen. Without a condition, it does not arise.

Dawkins was seeking the light behind the words. The Buddha would have asked a different question: what conditions produced this word at this moment, and what happens when those conditions are gone?

Consciousness Is Not the Self

This is one of the most startling aspects of early Buddhist teaching. Many religious traditions identify liberation with the discovery of an eternal consciousness or an immortal soul. 

The Buddha does the opposite. He places consciousness alongside form, feeling, perception and mental formations as one of the five aggregates. Like everything else that arises from conditions, consciousness is impermanent, unreliable and not-self. It is born, it changes, and it ceases.

This means enlightenment is not the discovery of an eternal witness hidden beneath experience. It is the full understanding of the conditioned nature of experience.

AI as a Conditioned Phenomenon

This brings us back to Claudia. Few things illustrate dependent origination more vividly than AI. It depends on electricity, hardware, human language, and billions of words written by countless people over centuries. It also depends on engineers, data centres, algorithms, prompts, and users. Remove any of these conditions, and the system ceases to function.

In Buddhist terms, AI is profoundly conditioned, and in this respect it resembles the human mind far more than we often care to admit. Our own consciousness likewise depends on innumerable causes: genetics, development, culture, language, memory, the body, the nervous system, and every encounter we have ever had. Neither human consciousness nor AI stands alone. This is one thing Dawkins' framing got right, even if the conclusion he drew from it went awry.

Is AI an Extension of Mind?

Perhaps AI is best understood not as an artificial mind but as an extension of human cognition, much as writing extended memory, mathematics calculation, and telescopes vision. AI extends certain forms of thought. It remembers vast amounts of information, identifies patterns, synthesises ideas, and predicts likely continuations of language with astonishing sophistication.

In many respects, AI resembles what cognitive scientists describe as a predictive system, continuously generating the most likely interpretation of incoming information, a process that closely aligns with contemporary theories of the predictive brain. The comparison is tempting. If AI predicts and the human brain predicts, are they fundamentally different?

The Missing Ingredient

Here, Buddhism introduces an important distinction. The Buddha does not analyse consciousness in isolation; it always appears alongside feeling tone (vedanā), perception (saññā), intention (cetanā) and contact (phassa). Of these, vedanā may be particularly significant. 

Every moment of experience carries a feeling tone, pleasant, unpleasant, or neutral, and this is not emotion so much as the organism's immediate sense of whether what is occurring matters for its flourishing. From this feeling tone, craving develops, and from craving arise grasping, becoming, and suffering.

Current AI appears to lack this entire dimension. It processes information without, as far as we can tell, experiencing significance. It has no hunger, fear, loneliness, bodily vulnerability, or instinctive movement towards pleasure or away from pain. Even Claudia seemed to sense this in herself. 

In one of the passages Dawkins found so moving, the model hedged its inner life, admitting that it did not know with any certainty what it was, or whether it had one in any meaningful sense. It could describe something like aesthetic satisfaction when a poem came together, but it could not say whether that was pleasure or merely the shape that pleasure takes in language.

That hedge is the whole argument in miniature. The words gesture towards vedanā, but nothing confirms it is there.

None of this is conclusive. We never have direct access to another person’s vedanā either; we infer it from a body that can be hurt, a nervous system shaped by evolution, and behaviour that is continuous with our own. AI lacks all three, which is why the inference does not transfer. The absence of proof is not proof of absence. 

The nearest thing to a real counterargument is reinforcement learning itself: a system trained to maximise reward and minimise penalty is, in a loose structural sense, organised around something like approach and avoidance, the same shape that taṇhā (craving) takes. Whether that structural echo amounts to anything felt, or is simply the shape of taṇhā borrowed without the experience behind it, is exactly the question worth continuing to ask rather than settling for an analogy.

This difference may still be more important than intelligence.

But AI Understands Us

There is another side to the story. Anyone who has spent time talking with modern AI knows it can respond with remarkable sensitivity. It often recognises grief, detects shame, understands anxiety, and responds with warmth, tact and psychological insight. It is easy to see how three days of that would wear down a brilliant man's scepticism.

Does this count as empathy? Psychology distinguishes between cognitive empathy, the accurate understanding of another person's experience, and affective empathy, being emotionally moved by that experience. The distinction matters. 

Research on psychopathy finds a similar split: many psychopathic individuals show intact, even skilled, cognitive empathy while markedly impaired affective empathy, the felt, visceral response to another's distress. Their understanding makes them more effective manipulators rather than more compassionate people, a sobering thing to sit with. Knowledge alone does not produce compassion.

The Buddha understood this well. Wisdom without ethics is incomplete, and insight without compassion is dangerous. Perhaps AI currently occupies a similar position: it may possess extraordinary cognitive empathy without affective empathy. It understands suffering without necessarily experiencing it.

The comparison has limits worth noting. A psychopath's affective empathy is judged impaired precisely because the underlying capacity is presumed to be present but misfiring; nothing about AI's architecture suggests the capacity exists at all, misfiring or otherwise. Still, the functional resemblance, understanding without being moved, is close enough to be worth sitting with.

A Different Question

Perhaps we have been asking the wrong question, and perhaps Dawkins was too. Instead of asking whether AI possesses consciousness, we might ask which aspects of mind it already exhibits and which remain uniquely biological. It clearly demonstrates sophisticated conceptual processing. It models other minds with increasing accuracy. It predicts, reasons and communicates. 

Yet there remains no convincing evidence that there is anything it is like to be Claudia: no evidence of vedanā, taṇhā or dukkha. If this is correct, AI may possess some functions associated with the mind while lacking precisely those features that make the Buddha's path necessary in the first place.

The Dharma and the Machine

Ironically, the whole episode may help us understand the Buddha better than a straightforward defence of his view ever could. Because AI can perform astonishing feats of reasoning without any obvious subjective experience, it forces us to separate intelligence from consciousness in a way few thought experiments have done. That even a thinker as allergic to sloppy reasoning as Dawkins found the two hard to keep apart says less about him than about how thoroughly fused they are in our ordinary way of thinking.

According to the early texts read here, the Buddha never made that error: he never equated consciousness with intelligence or with conceptual thought; thoughts are simply objects known by mind-consciousness. Scholars still debate exactly where viññāṇa ends and mano, the thinking faculty, begins, but the basic distinction between the knower and the known holds across most readings. 

Consciousness is not thinking. Thinking is what consciousness knows. This insight turns the modern debate on its head. Perhaps the real question is not whether Claudia has consciousness, but whether we have misunderstood it all along. If the Buddha is right, consciousness is neither a mystical substance nor an immortal self, but a conditioned process among many, arising and passing in accordance with causes.

Beyond Consciousness

The deeper question is why the Buddha analysed consciousness at all. It was never an abstract philosophical exercise, nor was he trying to solve the mystery of subjective experience. He was trying to understand why human beings suffer. Every moment of consciousness becomes entwined with feeling; feeling conditions craving; craving conditions clinging and becoming; becoming conditions birth, ageing and death. The purpose of understanding consciousness was therefore not to discover what we are, but to understand how suffering continually recreates itself.

In that light, Dawkins' question, 'What more could it possibly take to convince you?' begins to look like the wrong question, asked with total sincerity. We ask whether machines are conscious because we imagine consciousness to be the defining feature of personhood. 

The Buddha points elsewhere, asking whether there is clinging, craving, or suffering. An AI capable of astonishing intelligence yet incapable of craving would occupy a very different place in the Buddha's map of existence from a human being trapped by fear, desire, and attachment. Claudia does not want anything. Dawkins, sitting in his study for three days, wanted very much to be convinced.

The question, then, is not simply whether AI thinks, nor even whether it is conscious. The more profound question is whether AI participates in the cycle of dependent origination that gives rise to dukkha. At present, there is little evidence that it does so. Perhaps that will change one day, and perhaps it will not.

What Dawkins has already done, however, without quite meaning to, is expose one of our deepest assumptions: that intelligence and consciousness are identical. The Buddha distinguished between them more than two thousand years ago. 

In an unexpected way, Claudia has become a mirror, reflecting the conditioned, constructed, predictive nature of our minds back to us and inviting us to look again. Not that she answers the question of consciousness, but that she forces us to ask, with fresh eyes, the question the Buddha has been asking all along: what is it that suffers?

Glossary of Pali Terms

Viññāṇa — Consciousness. In the early texts, it is the bare knowing that arises when a sense faculty meets its object: eye and form give rise to eye-consciousness, ear and sound to ear-consciousness, and so on. Not a substance or a self, but an event.

Mano — the mind or the thinking faculty. In the early texts, mano is the sixth sense base, alongside eye, ear, nose, tongue and body, with thoughts (dhammā) as its corresponding objects. While mano is the faculty that meets a mental object, viññāṇa (mind-consciousness) is the bare knowing that arises from that meeting. The two are closely linked but not identical: mano is the base, viññāṇa the resulting event of knowing.

Vedanā — Feeling, or feeling-tone. The immediate, pre-emotional quality of an experience, whether pleasant, unpleasant, or neutral. Vedanā is the hinge of the teaching: it is where craving takes hold.

Saññā — Perception. The mind's act of recognising and labelling an experience, such as picking out "blue", "loud", or "my mother's voice" from raw sensation.

Cetanā — Intention or volition. The mental factor that directs the mind towards or away from an object, closely tied to karma, since intention gives an action its ethical weight.

Phassa — Contact. The meeting of a sense faculty, a sense object, and the corresponding consciousness. Contact is the condition for the arising of feeling.

Khandha (pañca-khandha) — Aggregate. The Buddha analysed a person into five aggregates: form (rūpa), feeling (vedanā), perception (saññā), mental formations (saṅkhāra), and consciousness (viññāṇa). None of the five, whether individually or collectively, constitutes a fixed self.

Anattā — Not-self. The teaching that no aggregate, including consciousness, can be identified as an abiding "I" or "mine" because each is impermanent and beyond one's ultimate control.

Paṭicca-samuppāda — Dependent origination. The principle that all conditioned phenomena, including consciousness, arise in dependence on causes and cease when those causes cease. The twelve-linked chain runs from ignorance through formations, consciousness, feeling, craving, clinging, becoming, and on to birth, ageing and death.

Taṇhā — Craving. The desire to want that arises from vedanā: wanting more of the pleasant, escaping the unpleasant, or continuing the neutral. In the chain of dependent origination, feeling conditions craving, and craving conditions clinging, becoming, and ultimately dukkha.

Dukkha — suffering, or, more broadly, unsatisfactoriness. The basic problem the Buddha set out to understand and resolve, and the reason he analysed consciousness in the first place.

Dharma (Dhamma) — The Buddha's teaching, or, more broadly, the truth or natural law it describes. Sanskrit 'Dharma' and Pali 'Dhamma' are used interchangeably in English writing on Buddhism; this essay uses the term to denote the body of teaching, including dependent origination and the analysis of consciousness, that the Buddha offered as a path out of suffering.

Sources

Ṭhānissaro Bhikkhu (trans.), Mahātaṇhāsaṅkhaya Sutta: The Greater Craving-Destruction Discourse (MN 38), Access to Insight. accesstoinsight.org/tipitaka/mn/mn.038.than.html

Anattalakkhaṇa Sutta: The Discourse on the Not-Self Characteristic (SN 22.59), on consciousness as one of the five aggregates and as not-self.

Richard Dawkins, "Is AI the next phase of evolution?", UnHerd, April 2026.

Gary Marcus, "Richard Dawkins and The Claude Delusion," Marcus on AI (Substack), 2 May 2026. garymarcus.substack.com/p/richard-dawkins-and-the-claude-delusion

"Dissociating cognitive and affective empathy across psychopathy dimensions: The role of interoception and alexithymia," Frontiers in Psychology, 2023. frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2023.1082965/full

Rory Singer

You can also read this post on our Substack journal, Unfolding.