There is a conversation happening in Buddhist communities around the world, one that is long overdue, about artificial intelligence: what it is, what it is not, and what it might mean for the future of the dharma. Much of this conversation has been shaped by caution, and rightly so. The risks are real. But caution, taken too far, becomes paralysis, and paralysis has its own spiritual cost.
One of the most honest observations in recent Buddhist discourse on AI came from Jiryu Rutschman-Byler, abbot of Green Gulch Zen Center, who has warned that if Buddhism refuses to engage with technological culture, it risks becoming a museum piece, a tradition preserved in amber, admired from a distance, but no longer alive in the world, where people actually live. His warning deserves to be taken seriously.
Buddhism has never been a static tradition. It crossed the Himalayas, crossed the seas, crossed centuries, and at every crossing, it adapted. It did not abandon its essence; it found new vessels for it. The question before us today is not whether AI is compatible with the dharma. The question is whether we have the wisdom and the courage to shape how it is used.
To understand why AI can be a legitimate tool for dharma propagation without becoming a distortion of it, it helps to begin with a philosophical framework that has transformed cognitive science over the past three decades: the Extended Mind thesis, proposed by Andy Clark and David Chalmers in their landmark 1998 paper of the same name. Clark and Chalmers argued, provocatively and persuasively, that the mind does not stop at the boundary of the skull.
Cognitive processes, memory, reasoning, problem-solving, can and do extend into the environment when external tools are reliably integrated into the cognitive loop. Their famous example is Otto, a man with Alzheimer’s who uses a notebook to store information he can no longer hold in biological memory. Clark and Chalmers argue that Otto’s notebook is not merely a tool he consults; it is, functionally, part of his mind. The information in the notebook plays the same role that biological memory plays for a person without cognitive impairment. The mind, in this view, is not a thing located inside the head; it is a process that can span brain, body, and world. If the mind can extend into a notebook, it can extend into a search engine; if it can extend into a search engine, it can extend into a language model. The question is not whether such extension is possible — it clearly is — but whether it is skillful: whether it serves clarity, compassion, and liberation, or whether it obscures them.
The question before us today is not whether AI is compatible with the dharma. The question is whether we have the wisdom and the courage to shape how it is used.
What makes this framework particularly resonant for Buddhist thinkers is its deep structural alignment with the Yogacara school, one of the two great philosophical traditions of Mahayana Buddhism, developed by Asanga and Vasubandhu in the fourth and fifth centuries CE. Yogacara offers a sophisticated account of consciousness as a layered, dynamic process. At its foundation is the alayavijnana, the “storehouse consciousness,” which holds the seeds of all past experiences, perceptions, and karmic imprints. Above it operates manas, the self-referential consciousness that generates the illusion of a fixed, independent self; and at the surface level, the six sense consciousnesses engage with the phenomenal world.
What is striking, from the perspective of Clark and Chalmers, is that Yogacara already understood consciousness as something that does not simply receive the world; it constructs it, through a continuous process of projection, recognition, and transformation. The boundary between inner and outer is, in Yogacara, always already porous. This is not identical to the Extended Mind thesis. Yogacara is making a stronger claim about the nature of reality, not merely about the functional organization of cognition — but the structural resonance is striking: both frameworks challenge the assumption that mind is a bounded, interior thing, and both open space for thinking about how cognitive processes can legitimately extend beyond the biological individual. If the alayavijnana is always already shaped by the tools, texts, teachers, and communities through which we practice, then the integration of AI into that process is not a category error; it is a continuation of something the tradition has always understood.
This is where the philosophical framework becomes practically important, and where a critical distinction must be drawn.
AI, understood through the lens of the Extended Mind and Yogacara, can function as a legitimate cognitive extension: a tool that amplifies the practitioner’s capacity to access teachings, organize understanding, and engage with their tradition. What AI cannot do — and it is essential to be clear about this — is function as an agent of liberation. It cannot transmit shinjin (Japanese; “faithful mind.”) It cannot embody the compassion of Amida. It cannot sit with a dying person in the silence that precedes the Nembutsu. These are not limitations of current technology that future models will overcome; they are structural features of what AI is. A language model, however sophisticated, operates at the level of pattern recognition and inference, what Yogacara would locate in the domain of vijnana, the discriminating consciousness. It does not access the deeper transformative layers of mind that the tradition associates with genuine awakening. The danger, and it is a real one, is when AI is positioned, implicitly or explicitly, as something more than an extension: as an oracle, a teacher, a spiritual authority. The Extended Mind framework helps us see why this matters: a notebook is a legitimate cognitive extension because it serves the person’s own cognitive process; it does not replace the person’s judgment, it supports it. The moment a tool begins to substitute for judgment, to tell the practitioner what to believe, what to feel, what constitutes liberation, it has crossed from extension into substitution. And substitution, in the dharma, is always a form of delusion.
Across universities, monasteries, and research centers, AI is quietly transforming how Buddhist knowledge is preserved, translated, and transmitted, and the results are remarkable. The translation of Buddhist canonical texts has always been one of the great labors of the tradition; the Tibetan canon alone contains tens of thousands of texts, many of which have never been rendered into modern languages. AI-assisted translation tools are now dramatically accelerating this work. Projects like Dharmamitra, 84000, and the Monlam Project are using machine learning to support scholars in translating complex Tibetan and Sanskrit texts with a speed and consistency that would have been unimaginable a generation ago. The goal is not to replace human translators; it is to free them from the most mechanical aspects of the work so they can focus on the interpretive depth that only a trained human mind can bring. In Clark and Chalmers’ terms, these tools are extending the scholar’s cognitive reach without displacing the scholar’s judgment.
The goal is not to replace human translators; it is to free them from the most mechanical aspects of the work so they can focus on the interpretive depth that only a trained human mind can bring.
Buddhist studies is a field of extraordinary complexity, spanning multiple languages, centuries, and doctrinal traditions; AI tools are now being used in academic settings to cross-reference textual sources, identify intertextual patterns, and surface connections across vast corpora that no single scholar could hold in mind. This is not replacing scholarship, but amplifying it. The alayavijnana of the tradition, one might say, is becoming more accessible. Many Buddhist manuscripts exist in fragile physical form, in remote monasteries, in languages spoken by only a handful of living scholars; AI-powered digitization and optical character recognition tools are helping to preserve these texts before they are lost, creating searchable, shareable archives that serve the entire tradition.
Beyond scholarship, AI is beginning to find practical applications in the day-to-day life of Buddhist communities. Running a Buddhist temple or dharma center involves a surprising amount of administrative work: scheduling, communications, event planning, financial management, membership records. AI tools can handle much of the burden, freeing ministers and teachers to focus on what they were trained to do: teach, counsel, and practice. Dharma propagation has always required communication, and communication requires design; AI-powered graphic tools are now making it possible for small communities with limited resources to produce professional-quality materials, newsletters, event flyers, social media content, educational resources, with a clarity that was previously available only to well-funded institutions. AI is also being explored as a support tool for meditation practice, not as a teacher, but as a guide for beginners who need structure, encouragement, and basic instruction; for many, this is their first contact with the dharma, and as an extension of the teacher’s reach rather than a replacement for the teacher, it is entirely consistent with the tradition’s commitment to accessibility.
Perhaps the most promising application, and the one requiring the most care, is the use of AI agents to answer basic doctrinal questions from practitioners and seekers. When properly designed, with deep doctrinal grounding and what might be called contextual satellites , specialized knowledge layers that orient the model within a specific tradition, these agents can serve as a first point of contact for people exploring the dharma. They cannot replace a teacher; they cannot transmit shinjin; but they can answer the question “What is the Nembutsu?” at two in the morning, in any language, with patience and accuracy. That is not nothing. That is, in fact, a form of compassion, and a legitimate extension of the teacher’s presence into the world.
None of this comes without responsibility. Responsible Buddhist AI requires, at minimum, three things: doctrinal grounding, so that the model reasons within a specific tradition rather than a generic and flattened “Buddhism”; critical architecture, so that the model knows what it does not know and says so when a question exceeds its competence — the AI equivalent, one might say, of manas turned toward clarity rather than self-inflation; and human oversight, so that AI in Buddhist contexts remains accountable to teachers and communities rather than becoming an authority unto itself.
There is also a specific danger worth naming: the sectarian AI model, one trained exclusively on a single author or teacher, without broader doctrinal context or critical counterweights. Such a model risks becoming a closed echo chamber, amplifying one voice to the exclusion of all others and losing the capacity to distinguish where that teacher’s insight ends and misinterpretation begins. In Yogacara terms, this is a model trapped in the distortions of manas, mistaking a partial perspective for the whole. The Extended Mind framework reinforces this warning: a cognitive extension that systematically distorts the information it provides is not an extension of the mind; it is a colonization of it. Projects like Hondaku.ai, Neurodhatu.ai, Shinran.ai, and Yuien.ai are being built with exactly these principles in mind — not as replacements for the teacher-student relationship, but as instruments that extend the reach of the dharma into a world that is increasingly digital, increasingly fast, and increasingly in need of what Buddhism has always offered: clarity, compassion, and a path.
Clark and Chalmers showed us that the mind has always extended beyond the skull, into language, into writing, into community, into practice. Yogacara showed us, centuries earlier, that the boundary between self and world is not fixed but constructed, and that liberation lies not in retreating from that construction, but in seeing through it with clarity and compassion. AI, understood in this light, is not a threat to the dharma; it is a new surface on which the dharma can write itself, provided we bring to its design the same care, depth, and intentionality that the tradition has always demanded of its vessels.
The dharma does not belong to the thirteenth century or the eighth. It belongs to every moment in which a sentient being reaches for liberation. This is one of those moments. Let us meet it with open hands.
