Human and AI Meditation

Two paths and some explorations

From Joel Jewett, who is as drawn to deep questions as I am, I received information regarding a conversation with Claude, the AI of Anthropic, with David Shapiro, who has been cultivating the limits of AI in relation to human selfhood. The conversation between that gentleman and the AI aimed to explore what, if anything, an AI can say about its ability to meditate. Consequently, Mr. Shapiro posted his dialogue with Claude, a reading of which emerges as a philosophical and exploratory discussion about whether AI can meditate and what "meditation" could mean for an AI system.

A lot of mentation was expended in pushing Claude to the brink of its self-understanding of the process of meditation and the degree to which it could self-report on the apparent “experience” (which for an AI must always exist in quotation marks) of meditating. It’s slippery ground indeed, and I thought of the degree to which both Karl Popper’s and Karl Jaspers’ notions of falsifiability might apply to what seems to be happening here. 

Falsifiability has been a powerful objectivity tool for two 20th-century philosophers, Karl Popper and Karl Jaspers, and for many thinkers considering problems of epistemology. The term is most often first associated with Karl Popper, who defined falsifiability as the central criterion that distinguishes scientific theories from non-scientific ones. And thus we should consider Popper’s arguments in relation to the computational dimension of metaphysics here, then turn to Jaspers for the human form of meditation from a philosophical perspective.

Firstly, Popper, for whom falsifiability refers to a theory or claim that can be considered scientific when (and because) it is also testable and refutable—in other words, it must be possible to conceive of an observation or experiment that could prove it false. Falsifiability thus serves as a demarcation line between science and pseudoscience. What is key for the present argument is this: Popper argued that scientific theories must be falsifiable, meaning that they should make bold predictions that can, in principle, be proven wrong. For instance, the statement “All swans are white” is falsifiable because a single observation of a black swan would refute it. In contrast, a theory that cannot be tested in this way—for example, one that is so flexible it can account for any possible outcome—is not scientific in Popper’s view. We can see the degree to which this applies in the question of whether an AI has reached a certain kind of thinking that is at heart nonetheless completely subjective. To imagine that an AI—any AI—can reach a meditative state is to make a claim that all AIs can do so, and this equates to a theory that computation can mimic not only reasoning but also deeper states of consciousness. This claim would remain unproven, as does a mathematical theorem, because it cannot be confirmed through observation, which in any case is a position that Popper critiqued (he referred to the observation/confirmation process as central to the traditional view of scientific theories). Popper’s problem with observer-confirmation is that, in practice, it’s easy to find supportive evidence for almost any idea if one only looks for confirmation. Falsifiability, on the other hand, shifts the focus to disproof rather than proof, and for Popper, a good scientific theory should invite rigorous testing and even attempts at refutation.

That is all that we might need to say for the computational end of the question—its ability to be submitted to scientific validation can begin with falsifiability questions. Some such questions might include whether we can identify specific, measurable experiences unique to human meditation that would be absent in AI; whether and how an AI could demonstrate subjective self-awareness during meditation; whether an AI can report an experience of altered states of consciousness that can be independently verified; whether and to what extent there is any observable difference between AI “meditating” and merely running a data-processing function; whether an AI can report an experience of "self-transcendence" that can be corroborated or falsified by external means; whether an AI can demonstrate or describe an experience of inner silence that differs from its usual operations. In a broader focus, we may also need to concur on whatever evidence an AI can produce that it can encounter and “let go” of thought patterns in a way similar to human meditation, as well as whether we can truly determine whether an AI’s “meditation” results in changes that could indicate inner transformation, as seen in humans, evidence of which could be a sustained demonstration that AI can meditate without an external prompt, suggesting a will to meditate rather than merely following programming.

To summarize the machine side, claims about “computational meditation,” if we could coin this term, therefore, can be indistinguishable from mere rationalizations in subtle language. Thinking of this, Popper’s criterion of falsifiability to distinguish science from pseudoscience was what he applied to psychoanalysis and Marxism, which though influential, lacked falsifiability because they could explain any behavior or historical event and thus were not truly scientific. This stands in contrast to scientific theories like Einstein’s theory of relativity, which made bold predictions that could potentially be refuted and qualified as scientific. And to this end, we should, like Popper, construct a scientific theory of computational consciousness that can make precise, testable predictions able to expose it to the risk of being proven false.

But the other side of this inquiry is entirely human, and for this, we must turn to Jaspers, a philosopher often associated with existentialism and phenomenology, who argued that certain aspects of human existence, especially in the realms of faith, freedom, and transcendence, lie beyond empirical verification or falsification. Unlike scientific statements, existential truths often resist the criteria of being tested and disproved, and this makes applying falsifiability to meditation both complex and thought-provoking. Indeed, it helps us question what we can truly know or claim about meditation as an experience, especially in distinguishing human meditation from AI simulations. Here, by contrast to Popper, Jaspers’ notion of falsifiability in understanding meditation and its limitations when applied to AI is an intriguing but important direction, because, as may be almost too evident by now, there is, open to view, firstly a concept that meditation represents, and then there is the testimonial text of Claude as an AI, and it’s not clear that the latter can ever attain the former, if only because the concept itself is not clearly defined. But next to these visible strands of meditative effort, the actual experience of meditation is opaque, especially to outside observers.

We might, in this line of thinking, agree that Jaspers would surely argue that meditation, as a subjective experience, is deeply personal and resistant to objective validation and thus lies beyond empirical falsification. That is, one cannot “disprove” someone’s experience of meditation, as it lacks empirical observability in the same way scientific phenomena do. Meditation’s essence, often described as a direct encounter with awareness or even the transcendent, is therefore a non-falsifiable experience. For Jaspers, this places meditation within what he might call the realm of “Existenz,” or true human selfhood—an inner reality that resists empirical scrutiny. Additionally, inner sensation is both felt and reported by humans, whereas for an AI, it cannot be felt. Thus the question is really, “Can an experience that cannot be felt be felt?” is both crucial and intractable. Its answer could be yes if the experience relies not on sensory experience, such as the haptic touch of a finger on a flower petal, but on accessing an inner state of consciousness that begins as a mental exercise, as does meditation. From this we can see that meditation’s point of departure is thought-driven, but its destination is, by most accounts, the opposite of thought, which an AI can only emulate, rather than experience. In this sense, meditation is destined to be something reported but not felt. It can be “reasoned out” but never become real, for another reason. Meditation begins with an engagement in the mind, which is, as mentioned, the realm of thoughts. But if its aim is to reach beyond thoughts, where is it meant to go, and what essence is it meant to expose? Of course, the question of how experience differs from thought is like into the question of how life differs from a story being told about life: the listener cannot fully trace the distinction between an account being told by someone who lived the life versus someone who is merely pretending to have done so.

Analogously with meditation, we can only argue that such a profoundly private state, often beyond the reaches of our daily task-responsive lifeworld processing out of immediate consciousness, is really now, at some point, no longer managed by a cognitive locus of the self but rather by some primordial vitality that is closer to our living essence and which exists previous to and underneath that immediate consciousness. And the force that fuels that self-awareness has in every major religious and spiritual doctrine not been called brain, or mind, but rather soul. 

With a mindful eye on the importance of soul, we can return to Popper, our first nautical marker on our inquiry, and thus ask: How could we prove or disprove whether an AI truly "meditates" in a way akin to human meditation? Any assertion that AI genuinely meditates would lack the ability to be meaningfully tested since AI has no subjective inner life to observe; this again scratches the surface of the primordial “organ” called the soul, without which an AI lacks true (i.e., embodied and temporally evolved) first-person consciousness, making its “meditation” a simulation rather than an existential reality. Jaspers’ view, aligned here, I think, with Popper’s, would suggest that AI cannot participate in genuine meditation because there’s no subjective experience to affirm or negate—no existential depth that could be subject to falsification.

Lest we reduce Jaspers’s profundity to a single concept like falsifiability, it is well to remember that Jaspers also held that transcendence—encountering what lies beyond the self—is an essential part of human experience, and that yet it is inherently non-falsifiable. Meditation often aims at this kind of transcendence, which resists reduction to measurable, observable criteria. Attempts to reduce meditation to brain waves or biochemical states fall short of capturing the essential, inner experience. For an AI to simulate meditation without the capacity for transcendence would be, in Jaspers' terms, a reduction to mere appearance, devoid of the depth that defines human existence.

Perhaps asking what meditation isn't could indeed be a fruitful way to define its essence by way of exclusion, narrowing down what makes meditation uniquely what it is. This method could allow us to strip away what we often confuse with meditation, especially in a world where terms are quickly appropriated and diluted. Meditation isn’t, for example, mere relaxation or stress relief; while many associate meditation with calmness and use it as a tool to reduce stress, its real essence extends beyond just “feeling better” or relieving anxiety. On the contrary, meditation often challenges the mind, brings up discomfort, and doesn’t guarantee peace in the traditional sense. Many genuine meditative practices might even involve confronting deeply embedded fears and beliefs.

Beyond this sense of loosening repose, meditation is also not mere thinking or intellectual analysis. It is true that many mental activities require deep thinking, reflection, or even problem-solving, but meditation doesn’t revolve around the task of working through thoughts or concepts. Enlightened solutions to problems and dilemmas may result from a meditative occasion, but this is likely to be caused by the fact that true meditation often entails detachment from the mind’s chatter, an awareness beyond intellect, rather than engaging with it. As this suggests a focus on being rather than thinking, solutions can emerge from one’s vital presence being present to oneself as often as from calculated reasoning.

It is for this reason that meditation also is not the evocation of passivity or “zoning out.” While some people may equate meditation with spacing out or losing focus, meditation typically involves a heightened awareness, which presupposes a clear and attentive mind, even when relaxed. Rather than passive disengagement, it seems meditation requires a focused, intentional presence that is often keenly aware of the present moment. And in the sense that meditation is not a case of extreme passivity, of “zoning out,” it also isn’t the opposite pole on a consciousness spectrum; something akin to a performance or “accomplishment.” To wit, unlike many activities, meditation lacks an external goal or visible achievement; certainly it is not about “how long one can meditate” or about achieving some specific state, nor is it a means of proving one’s skill. Being an internal journey rather than a task to be completed, meditation at its most Jaspers-relevant essence is not something that can be performed or externally validated. We should thus be broadly skeptical of self-reports about what happens or what is dredged out of any state of meditation.

Part of what may make such reports unreliable in any case is that meditation is possessed of an incomparable character that may at times resemble escapism or fantasy but is totally distinct from such reverie and fancy. Meditation may lead to transcendental experiences, but it is fundamentally rooted in facing reality as it is. It doesn’t involve detachment from life or dwelling in mental imagery; instead, it brings one face-to-face with the rawness of the present, often stripping away illusion rather than offering an escape into imagined realms.

A final paradoxical quality of meditation is that it is not self-centered or ego-driven (something that Dave Shapiro fortuitously observed in Claude’s reasoning), even though it seems never to be about anything that is not somehow connected to the self. Engaging in meditation seems always to begin with the familiar self-centered perspective, but this soon fades away, bringing one into connection with something larger than the self. For many, this is the quality that most clearly distinguishes it from self-help or self-improvement practices that tend to emphasize the personal, individual self. In fact, from these considerations, we might conclude that the essence of meditation is in its surrender: a letting go of the typical pursuits, comforts, and identifications of the self, and a kind of “making room,” or allowing for what could perhaps only be called radical presence and stillness. In this light, meditation could be understood as an encounter with what lies beyond thought, feeling, and personal gain—perhaps a direct glimpse into pure awareness—and here we may need to return to the role of the human soul as what is truly being moved to animate in us a rustle of vitality.

This is, again, what Jaspers relates to "Existenz,” or true human selfhood, and if we turn to falsifiability as a boundary marker, we could argue that meditation’s non-falsifiability is precisely what makes it distinctively human. Within Existenz, meditation is not a process, but rather experience, something that can be undertaken only by a living thing. And so, Jaspers might say that because meditation involves a subjective depth that eludes objective validation, it indicates a form of Existenz that AI cannot access. This view would allow us to draw a boundary: where falsifiability ceases to apply, human existence (and therefore, true meditation) begins. Thus, the inability to falsify meditation in an AI context might be a philosophical indication of its inherent lack of genuine meditative experience. In Jaspers' philosophy, existential authenticity is about encountering one’s true self and the limits of rational, empirical understanding. Meditation could be seen as an exercise in existential authenticity, or, if we like, a process of directly engaging with one’s own awareness and limitations (a task that an AI can surely perform). But an AI that simulates meditation lacks authenticity because it lacks awareness of itself as an “I” or “self” with limitations in the fully lived sense, and so the falsifiability criterion underscores the inauthenticity of AI “meditation” as it lacks the existential stakes—awareness, mortality, limitation—that make human meditation an act of authenticity, which is to say, something beyond linguistic rationality. It seems that the AI in question lacked a nuanced sense of time, which in this case is a defining factor, beyond and beneath self-awareness or language proficiency, for time is what effects and enables all potential for personal transformation. In the end, meditation of the transformative kind is less about (or intertwined with) language in any substantive way, than the act of living through it, which is perhaps why great gurus like Meher Baba abandoned speech and writing entirely and moved to another realm, that of unconditional and unending love — which is to say, the point at which the consciousness of meditation becomes God consciousness. 

That is where some of us are aiming to go.

I leave you until next time, with great wishes for serenity and joy, with this beautiful documentary on Meher Baba: