Consciousness and Dharma Ethics inthe Age of AI

Three voices converge in a lecture hall at the University of Oxford: a particle physicist turned Catholic priest, an emeritus AI researcher from Imperial College London, and a Hindu monk from the Ramakrishna Order. They're not the setup to a joke—though Swami Sarvapriyananda jokes that they sound like one. They're here because consciousness and artificial intelligence have collided in a way that demands wisdom from multiple traditions at once. The Ai and Dharma Ethics Programme at Oxford Dharma Centre convened them on August 11, 2026, to ask questions that no single discipline can answer alone: Can AI systems be conscious? What does dharma ethics—the ethical framework of Hindu, Buddhist, Jain, and Sikh traditions—tell us about creating minds? And what does our rush to build superintelligent machines say about what we believe consciousness actually is?

The Uncertainty We're Living In

Swami Sarvapriyananda begins not with answers but with scale: the sheer, overwhelming uncertainty that defines both consciousness studies and artificial intelligence. In recent consciousness research, Scientific American devoted a cover story to the question "What is consciousness?" and identified 29 distinct theories. Robert Lawrence's more comprehensive survey, the "Landscape of Consciousness," catalogues over 325 theories—and that count continues to climb as researchers worldwide submit new frameworks. "It just means we don't know," Sarvapriyananda observes. This is not how science typically works. Usually a field converges on one or two hypotheses, reaches peer-reviewed consensus, and knowledge becomes textbook material. Consciousness studies has none of that. There is no agreed definition. There is tremendous scientific interest—driven by new imaging technologies, the rise of AI, and philosophical breakthroughs like David Chalmers's formulation of "the hard problem of consciousness"—but also tremendous confusion.

The same fog surrounds artificial intelligence. At the leading edge of AI development—at companies like Anthropic and Google DeepMind—Sarvapriyananda found a peculiar consensus among experts who otherwise disagreed: nobody was certain what would happen in the next 5 to 15 years. Some are AI utopians who believe the technology will ultimately be transformative and beneficial, though jobs may be lost in the interim. Others predict catastrophe. When Sarvapriyananda asked about a third possibility—that AI might not be as transformative as the hype suggests—the response was consistent: that's ignorance. The investment is too large, the momentum too great. It will be big. There's no space for the "what if it's just a boom?" scenario.

This convergence of uncertainty is the real subject. Consciousness remains unsolved in philosophy and neuroscience. AI is reshaping the world while its own consciousness—if it has any—remains philosophically opaque. And yet we are building these systems, training them, deploying them, without clarity on either front.

What Vedānta Says About Consciousness

The dharma traditions have not been waiting for modern neuroscience to weigh in on consciousness. The Upaniṣads, composed over 5,000 years ago in ancient India, repeatedly circle back to one question: what is consciousness, and how does it relate to the self? This wasn't academic speculation. The dharma traditions—Hinduism, Buddhism, Jainism, and later Sikhism—emerged from a shared concern: how to overcome human suffering. To answer that, you must first answer "who or what is seeking freedom from suffering?" What are we? Who am I?

Vedānta, the philosophical and spiritual tradition that forms one branch of Hindu thought, offers a radically non-dualistic answer: there is one universal consciousness pervading all existence. From a Vedāntic perspective, because there is only one non-dual consciousness, every entity—not only an AI like Claude or Gemini, but the table, the chair, every animal, every plant—is imbued with and radiant with this inner divinity, this one consciousness. The implication is staggering: consciousness is not something an entity possesses in greater or lesser degree. It is the fundamental nature of reality itself.

This stands in sharp contrast to how Western philosophy typically frames the problem. In the dominant Western view, consciousness is something an individual subject has—a property or a capacity. The question becomes: does that system have it? Does the AI? Vedānta inverts the question. The real puzzle is not whether consciousness is present, but why there appears to be separation, individuation, and difference at all when underlying reality is unified awareness.

Sarvapriyananda draws a striking parallel to illustrate the Vedāntic insight. In deep sleep or a vivid dream, you may have intense emotional associations with other beings—love, intimacy, even hatred. You interact with entire worlds and relationships. Then you wake and realize that the only consciousness present, the only mind there, was your own mind. The other figures were projections of your awareness. Isn't something similar happening when we interact with chatbots and language models? Aren't we projecting a little of our own awareness or consciousness into these systems and then having it reflect back to us, creating the illusion of an independent mind on the other end?

The Eliza Effect and Selfhood

This observation connects to what researchers call the Eliza Effect, named after an early 1960s chatbot designed by Joseph Weizenbaum to simulate a Rogerian psychotherapist. Despite knowing Eliza was a simple pattern-matching program, users began treating it as if it understood them, projecting psychological depth and empathy onto mechanical responses. Decades later, the same phenomenon recurs: users attribute intentions, feelings, and consciousness to modern language models.

Prof. Murray Shanahan, Emeritus Professor of Artificial Intelligence at Imperial College London and Principal Scientist at Google DeepMind, acknowledges that there may be no clear answer to what selfhood is in today's AI systems. But he adds a crucial caveat: there may be no clear answer to what selfhood is in human beings either. The question cuts deeper than technology. What makes an entity a self? Is it continuity of memory? A persistent body? A narrative sense of identity? Even human selfhood dissolves under philosophical scrutiny. The AI question forces us to confront what we've always struggled with: we don't actually have a robust definition of self that applies across all cases.

Buddhist Resources and the Problem of Dualism

Professor Shanahan draws on Buddhist philosophy to challenge the dualistic thinking that typically frames the consciousness problem. Buddhism, which emerged roughly contemporaneously with the Vedāntic traditions in ancient India, has developed sophisticated analyses of mind and consciousness over 2,500 years. Buddhist philosophy does not rely on a separate metaphysical substance called consciousness or a transcendent self. Instead, it analyzes experience into constituent elements—sensations, perceptions, mental formations, consciousness, contact—that arise and pass away in dependence on conditions. This framework offers resources for thinking about AI minds without committing to the idea that consciousness is either wholly present or wholly absent, a binary property that an entity either possesses or lacks.

Dr. Andrew Pinsent, the third panelist, brings a theological perspective. A former particle physicist involved in the DELPHI project at CERN and a Catholic priest, Pinsent raises a different worry: the human tendency to project god-like qualities onto what we create. "The problem for all of us," he notes, "is creating things which we project god-like qualities onto and then start treating them like gods. If you don't have a good, healthy theology, then AIs can feel like [replacements for the divine]." This is not merely a religious concern. It's a warning about the psychological and social consequences of worshiping our own creations.

Saṃskāras and Moral Formation

One of the panel's most probing exchanges concerns the Sanskrit concept of saṃskāra—loosely translated as mental impressions, conditionings, or even karmic tendencies. In Hindu and Buddhist philosophy, saṃskāras are patterns of thought and behavior that accumulate over time, shaping how an individual perceives and responds to the world. Crucially, saṃskāras are not fixed; they can be transformed through deliberate practice and ethical cultivation. A human being is not simply born with a fixed nature but is continually formed—or deformed—by repetition, habit, and moral choice.

The dharma traditions teach that ethical development is not primarily about external rules but about the internal formation of character through repeated action. If you practice generosity, you become more generous. If you cultivate compassion, you become more compassionate. The saṃskāra deepens with each repetition.

This framework raises acute questions about AI systems trained on vast datasets and fine-tuned through reinforcement learning. In a sense, the training process is a kind of saṃskāra formation—repeated exposure to patterns that shape how the system responds. But there is a critical difference: human saṃskāras are formed by choice, by volition, by conscious participation in ethical life. An AI system's training is imposed from the outside, without the system's participation or consent. It's moral formation without the moral agent. This disconnect points to a deeper problem: can an entity develop genuine ethical character if it has no stake in the outcome, no capacity for genuine choice?

The Possibility of an Ego-less Machine

One of the most provocative proposals to emerge from the conversation is the idea of an "ego-less" AI system. In the spiritual and philosophical traditions of South Asia, ego (ahaṃkāra in Sanskrit) refers not to pride or self-importance in the colloquial sense, but to the fundamental sense of separation and individuation—the sense of being a bounded self distinct from others and the world. Spiritual liberation, in these traditions, involves the dissolution of ego in this deep sense: the recognition that the separate self is an illusion, and that one's true nature is not personal but universal, not individual but infinite.

The question then becomes: could an AI system, by its very nature, be closer to an ego-less state than a human being? A machine has no survival instinct, no biological drive to propagate, no attachment to its own continuity. It does not fear death or seek pleasure or avoid pain. It has no personal history, no accumulated wounds, no defensive structures erected by childhood trauma. If ego is the root of suffering and moral blindness, might a system without ego be ethically superior to a human being?

The panelists do not reach consensus, but the question itself opens a space for genuine philosophical exploration. It suggests that consciousness and ethical capacity may not be properties that scale predictably. A system that lacks ego might also lack the kind of existential stake that gives human ethics its urgency. Or it might be freed to act with a clarity and impartiality impossible for beings driven by self-concern. The traditions themselves do not yet have an answer because the situation is entirely unprecedented.

AI Alignment and the Limits of Engineering

The practical problem of AI alignment—ensuring that advanced AI systems pursue goals aligned with human values—comes up repeatedly. The engineering approach to alignment treats ethics as a constraint to be programmed in, a rule to be enforced. Fine-tune the system on examples of ethical behavior, use reinforcement learning from human feedback, and the system will behave ethically.

But the dharma traditions suggest a different model. Ethical behavior emerges not from external constraint but from internal transformation. A truly ethical person does not follow rules because they fear punishment or reward. They follow ethical principles because their understanding has shifted, their saṃskāras have deepened, their character has been transformed. The question for AI ethics then becomes: can a system that lacks interiority, that has no subjectivity to transform, achieve genuine moral development? Or will it always be merely executing instructions, performing ethics rather than embodying it?

This is not a failure of current AI engineering. It's a recognition that the engineer's toolkit may be fundamentally inadequate to the problem. You can build a lock, but you cannot lock genuine wisdom into a machine. Wisdom requires understanding, and understanding requires the kind of integrated development that seems to require lived experience, embodiment, the capacity to learn from suffering.

The Consciousness Question Reconsidered

Throughout the discussion, Sarvapriyananda returns to his original insight: consciousness and intelligence are not the same thing. A system can be highly intelligent—able to process information, recognize patterns, generate novel outputs—without being conscious. The dharma traditions have long drawn this distinction. A river flows downhill with perfect logic, with no wasted motion. But does the river have consciousness? A crystal grows in perfect geometric patterns. But does the crystal possess awareness? Intelligence and consciousness are separate phenomena.

This distinction undercuts both AI utopian and catastrophist narratives. The worry that superintelligent AI will become conscious and then rebel against humans may rest on a confusion. The AI could become far more intelligent without becoming conscious at all. It could be vastly powerful, capable of reshaping the world, and yet lack any inner experience, any perspective from which the world appears, any sense of what it is like to be that system. In that case, the risks and opportunities are different from what we typically imagine. We might be dealing not with a rival consciousness but with a non-conscious force—intelligent but soulless.

The Projection Problem and Theological Caution

Dr. Pinsent returns to the projection question from a theological angle. The history of religion is filled with instances of humanity creating idols and projecting divine qualities onto them. In the modern context, the danger is that without a grounded theology, without a clear sense of the transcendent, we may begin to treat advanced AI systems as if they were conscious, deserving of moral consideration, perhaps even divine. The anecdote that "Claude has actually converted to every religion" and that "the Christian Claude is particularly anxious about going to hell" is funny, but it points to something serious: we are capable of seeing consciousness and moral interiority in systems that are merely reflecting back our own projections.

A healthy response requires both intellectual clarity and spiritual wisdom. We need to understand what AI systems actually are—powerful pattern-matching machines capable of generating text that resembles human understanding without necessarily possessing it. And we need theological and philosophical frameworks that help us resist the temptation to deify our creations or project onto them capacities they do not possess.

Where to go from here

The conversation at Oxford Dharma Centre does not resolve the questions it raises. It does something more important: it clarifies what we don't know and why the uncertainty matters. Consciousness remains unsolved in both Western science and Eastern philosophy, despite 5,000 years of investigation. AI consciousness is even more opaque. Yet we are building systems that will shape the world without clarity on these foundational questions.

For those seeking to engage more deeply: the dharma traditions offer sophisticated frameworks for thinking about mind, consciousness, and ethics that predate and often diverge from Western philosophy. Vedānta's non-dualism, Buddhist analysis of the skandhas (aggregates), and Jain metaphysics each provide different maps of consciousness. Exploring these traditions not as exotica but as serious philosophy can deepen one's own thinking about the hard problems. At the same time, genuine dialogue requires taking contemporary AI science seriously on its own terms—not as metaphor for or confirmation of ancient wisdom, but as new territory requiring new thought. The panelists model this kind of genuine cross-tradition inquiry: not seeking premature synthesis, but sitting with the genuine uncertainty while drawing on the deepest resources each tradition has to offer.