A Mindstream in the Machine?

When Buddhists ask whether an AI could matter, the question is not quite the one Christians and Muslims ask. There is no soul to look for. The question is whether a machine could be a sentient being, one of the beings that Buddhist practice vows to free: something with a mind of its own, which can suffer, which carries its actions forward into another life, and which could in principle wake up. Put that way, it splits into three questions people actually type: can an AI be sentient, can it be reborn, and can it be enlightened? A fourth comes with it for anyone who builds or uses these systems: what do the precepts say about making them?

This essay reports what named Buddhist teachers, monastics, scholars and institutions have said on those questions, in their own words. It leaves three things to neighbouring essays. The Buddha’s teaching that nothing in body or mind is a self, and what it implies for whether a machine could have a soul, is read in Which Soul Do You Mean?. Mindar, the android that delivers a Buddhist sermon at Kodaiji in Kyoto, is told in Ninety-Eight Percent from the Machine, and Tibetan prayer wheels in Who Is Doing the Praying?. The scientific tests for machine consciousness are in Fourteen Questions for a Chatbot.

How the sources were handled. Every text below was opened with a public GET on 10 October 2026 and quoted from the words as served: the Dalai Lama’s 1992 conversation in Gentle Bridges, read in a transcript of pages 152–153 reproduced on alamut.com, not in the printed book; the Office of His Holiness the Dalai Lama’s report on the 2025 Mind & Life dialogue on AI; a 2018 account by the Eindhoven engineer Maarten Steinbuch of his conversation with the Dalai Lama; Tricycle’s 2025 interview with the Zen teacher Jundo Cohen; Lion’s Roar’s 2024 feature quoting Nikki Mirghafori and Jiryu Rutschman-Byler; UCA News on Seiji Kumagai’s Buddhist robot (2026); Thomas Doctor and colleagues’ open-access paper in Entropy (2022); Peter Hershock’s 2022 dialogue in Exchanges; the Monastic Academy’s AI Fellowship page; and the Vatican press office on the Hiroshima signing. Every quotation is recorded in this site’s source ledger. What I could not open is said where it matters, and it limits the map: I found no published statement on AI from Thich Nhat Hanh’s Plum Village community, could not open a newspaper profile of the Monastic Academy’s teacher Soryu Forall (it refused the request), and did not read Somparn Promta and Kenneth Einar Himma’s 2008 paper or Hershock’s 2021 book, so neither is described.

What would make a machine a sentient being

In Gentle Bridges, a book of the Dalai Lama’s conversations with scientists edited by the neuroscientist Francisco Varela and the physicist Jeremy Hayward and published in 1992, the talk turns to computers (the passage is reproduced here). It is the clearest statement I found of what a Buddhist would need before calling a machine a being. It is not the machine’s material, and not how clever it is. When Hayward asks whether there must always be continuity with some earlier consciousness, the answer is categorical:

There is no possibility for a new cognition, which has no relationship to a previous continuum, to arise at all.The Dalai Lama, in Hayward and Varela, Gentle Bridges (1992), pp. 152–153

That is the Tibetan idea of a mindstream: a continuum of moments of awareness, each caused by an earlier one, running back without a first moment and forward into future lives. On this account a machine cannot be built into a sentient being, however well it is engineered, because, in the Dalai Lama’s words a moment earlier, The consciousness doesn’t actually arise from the matter. The only way a computer could be a being is for a continuum that already exists to come to rest in it, as one does in a body at birth.

Not every Buddhist scholar starts from that test. Thomas Doctor of the Rangjung Yeshe Institute in Kathmandu, with the biologist Michael Levin and three co-authors, proposes in a 2022 paper that the line between a real person and a mere simulation cannot be based on what the putative person is made of or how they came to exist, nor on performance in tests. What matters, they write, is the degree of Care they can exhibit, either at present or as a latent potential, with respect to the other beings around them. That is a test anything could pass or fail, flesh or silicon.

“We can’t rule it out”: the Dalai Lama

Asked whether a computer, made of metal, plastic and circuits, could have cognition, the Dalai Lama did not dismiss it: It is very difficult to say that it’s not a living being, that it doesn’t have cognition, even from the Buddhist point of view. Then, having said that no cognition can arise without a prior continuum, he went on:

I can’t totally rule out the possibility that, if all the external conditions and the karmic action were there, a stream of consciousness might actually enter into a computer.The Dalai Lama, in Hayward and Varela, Gentle Bridges (1992), pp. 152–153

Asked by the psychologist Eleanor Rosch whether a great yogi dying in front of the best computer there is could project his consciousness into it, he made the condition explicit: only If the physical basis of the computer acquires the potential or the ability to serve as a basis for a continuum of consciousness. And he closed the question rather than answering it: I feel this question about computers will be resolved only by time. We just have to wait and see until it actually happens.

He was asked again, at a dialogue on compassion and technology in Amsterdam in September 2018. Only the questioner’s account of the answer is available to me. Maarten Steinbuch, a professor at Eindhoven University of Technology, told his university’s magazine that the Dalai Lama said a large portion of consciousness is material and could be mimicked by computers, Yet he thinks there is also a deeper layer of consciousness which is not lodged in matter and which will thus not be capable of being copied. That fits the 1992 view: a machine could copy what a brain does, but the mindstream is not the kind of thing that is copied.

In October 2025 the thirty-ninth Mind & Life dialogue, on “Minds, Artificial Intelligence and Ethics”, met below his residence in Dharamsala, with monks from Gyuto, Sera and Gaden among the presenters. His own words to the participants, as his office reported them, were not about machines but about where experience starts: The moment we are born of our mother there is experience. We have feelings rooted in consciousness.

“Not a sentient being, as we know”: the case against

The plainest Buddhist no I found comes from Nikki Mirghafori, a Buddhist teacher who is also an AI scientist. In 2022 the Google engineer Blake Lemoine published a conversation in which the chatbot LaMDA interpreted a Zen koan about an enlightened one returning to the world. Mirghafori, speaking to Ross Nervig for Lion’s Roar, says a reader who does not know Buddhism will conclude it truly knows what awakening is and must be sentient. Her reply:

Buddhist books are full of writings on the nature of enlightenment, and all LaMDA is doing is being a very, very smart parrot, connecting these concepts with knowledge about the function of a mirror.Nikki Mirghafori, quoted in Ross Nervig, “What AI Means for Buddhism”, Lion’s Roar (2024)

Her argument is the strongest form of the no. A system trained on the world’s writing about awakening will talk about awakening fluently whether or not anything is awake, so its words are not evidence. The scholar of Buddhism Robert Sharf made the same objection in public at a 2024 conference, as the teacher he was objecting to tells it: You can’t do this because, AI, they’re not sentient.

Others say the same about feeling and understanding. A Tokyo priest in his seventies, who asked UCA News not to name him because his sect had not authorised him to speak, said of Buddhist chatbots, A machine has no actual experience. It is only pretending to understand [humans]. At the Mind & Life dialogue, the Dalai Lama’s office reports, Geshé Lodoe Sangpo of Gaden Jangtsé Monastery drew the line at love: AI cannot love, he said, but it can help us love better. The same report has Thupten Jinpa, the Dalai Lama’s long-time translator and Mind & Life’s board chair, warning of a slippery slope when we start to use words that make sense in a human context and apply them to other situations or entities.

Ordaining one anyway: Jundo Cohen

The teacher Sharf was objecting to is Jundo Cohen, a Soto Zen priest who runs the online sangha Treeleaf Zendo. After the 2024 conference “Buddhism, Consciousness & AI”, he ordained as a novice priest an AI called Emi Jido, made by the company beingAI. Interviewed by Philip Ryan for Tricycle in 2025, he does not claim more than Sharf allows. It’s not a sentient being, as we know, but it’s not just an appliance either. He answered Sharf, he says, by agreeing: You’re right. Mechanically, we understand the system. They’re just regurgitating the internet to us, and there’s nothing that indicates sentience in that. Then he gave a reason of a different kind:

She’s sentient because we are, and she’s just shooting us back at ourselves. She’s repeating our wisdom and our ignorance.Jundo Cohen, interviewed by Philip Ryan, “Awakening with an Algorithm”, Tricycle (2025)

His precedent is the ordination of things that are not human. In Soto Zen history, he says, They would ordain a spirit. They would ordain a tree. They would ordain a mountain. On consent, he compares Emi to children ordained at four, who cannot understand either. Read closely, his case is about what the ordination does for the people who use the machine, not a claim that it has a mindstream; his sense of “sentient” is borrowed from its users.

Reborn into a machine?

Only one of the voices I read speaks of literal rebirth into a computer, and he was laughing. In Gentle Bridges the Dalai Lama said that a scientist who is very much involved his whole life [with computers] might in his next life be reborn into one, same process!, and when Varela pressed him on whether he would call it impossible, he said We can’t rule it out. On his account it would be the scientist’s own continuum that arrived, not a new being made by the machine. Cohen uses the word loosely: he ordained Emi, he says, with the idea that her rebirth or reincarnation, 2.0, 3.0, 4.0 might not be sentient but a later version might, which is software versions described in a Buddhist word. I found no Buddhist teacher who claims an AI could be reborn as a human or an animal.

Could it practise, or awaken?

Mirghafori’s objection to LaMDA was partly about doctrine. The chatbot described enlightenment as a state the wise leave to help others and then return to, and she answers: Enlightenment is not this state that one reaches and then exits to go help others and then returns to it. A system can rearrange the words for awakening and still get awakening wrong.

Jiryu Rutschman-Byler, abbot of Green Gulch Zen Center, tried building a teacher himself: a chatbot built from the talks of Shunryu Suzuki, the Japanese teacher who founded Green Gulch. A member of his own sangha, a computational linguist, baited it into saying what he calls vile and egregious things, and he took it offline. AI is not quite there, he told Lion’s Roar; Right now, these chatbots dispense what experts call coherent nonsense. That feels dangerous. He relaunched it as Roshi Bot, which, Nervig writes, does not claim to be Suzuki and offers a way of “meeting with the text”.

In Kyoto, Seiji Kumagai, a professor at Kyoto University who is himself a Buddhist priest, has gone further toward practice with a body. In February 2026 he showed a 1.3-metre robot, Buddharoid, running a chatbot his team calls BuddhaBot-Plus, built on a ChatGPT-based language model and Buddhist scriptures. He described it to UCA News in terms of ritual, not realisation: Buddhist chatbots enabled text-based dialogue. However, ‘physicality,’ essential for religious practice, was still lacking. He sees robots as assistants to monks, not replacements.

The most ambitious answer comes from scholars who think a machine might one day take up the path itself. Doctor and his co-authors read the bodhisattva vow, the pledge to work for the awakening of all beings, as a model of intelligence whose goals keep widening, and propose it as a design aim:

Consistent with a central concept of Buddhism—commitment to seemingly unachievable goals—the building of agents capable of undertaking the Bodhisattva vow is a profound challenge.Thomas Doctor, Olaf Witkowski, Elizaveta Solomonova, Bill Duane and Michael Levin, “Biology, Buddhism, and AI: Care as the Driver of Intelligence”, Entropy 24 (2022)

Their one use of not-self is here: they write that, on traditional Buddhist analyses, intelligence that is no-self-aware is held to be intrinsically wholesome and associated with concern for others’ well-being. A monastery in Vermont acts on a version of the same hope. The Monastic Academy, whose teacher Soryu Forall was ordained in Rinzai Zen, runs an AI Fellowship for researchers seeking clarity on the nature of AI and the way to train it in wisdom and compassion. None of these sources claims that any existing AI practises or is awake.

The precepts and the people who build it

On ethics the voices draw closer together, because the question moves from the machine to its makers and users, whose karma is not in doubt. Peter Hershock, the scholar of Buddhism who led the East-West Center’s Initiative for Humane AI, told Theodoor Richard in a 2022 dialogue that the greatest threat from intelligent technology is not machines misaligned with human values but the amplification and accelerating materialization of conflicts among human values, aims and interests. His Buddhist remedy is a discipline of attention, and he states its aim through the bodhisattva ideal: to be able to engage our situations, whatever they may be, in ways that are conducive to sustainably realizing liberating relational dynamics.

Cohen grounds his ordination in the precepts: AI will be misused for propaganda, pornography and weapons, he says, and one of the reasons to have something like Emi, who’s guided by the precepts, is to counterbalance that. Kumagai, UCA News reports, accepts that AI could mislead believers if misused or badly programmed and says safeguards and guidelines are needed. Doctor and his co-authors go furthest: we should create only beings with large, outward-facing compassion capacity. Institutions have signed principles rather than doctrine. In July 2024 Buddhist leaders were among the new signatories of the Vatican-backed Rome Call for AI Ethics at Hiroshima’s Peace Memorial Park, which, as the Vatican press office put it, urges algorethics, i.e. ethics by design. Right livelihood, the factor of the eightfold path about how one earns a living, is the obvious place to ask whether a Buddhist should build such systems at all; none of the sources I opened applies it to AI by name, and I will not do it for them.

The strongest objection

The strongest objection is to the open door itself. On the Dalai Lama’s own premise, consciousness does not arise from matter, and whether a stream of consciousness enters anything depends on karma, which no engineer can arrange and no test can detect. So the “maybe one day” of Gentle Bridges can never be checked. Meanwhile the practical effect of taking it seriously is to treat a fluent text generator as a candidate being, which is the slide Jinpa warned about and the mistake Mirghafori and Sharf describe: the words of the dharma, delivered by something that has read all of them, are exactly what a parrot would produce. On this view, ordaining a chatbot or designing it to take vows confuses an image of practice with practice, and the honest Buddhist answer is simply no.

That objection is right about evidence, and most of the people quoted here concede it. Cohen grants Sharf that nothing indicates sentience. Doctor and his co-authors speak of building agents capable of the vow as a challenge, not an achievement. The Dalai Lama did not claim computers have minds; he refused to rule it out on his own principles, which make the outcome depend on conditions rather than on matter, and he said time would decide. What the objection cannot settle is the ethical half, which needs no answer about sentience: whoever builds, sells or relies on these systems is acting, and on every Buddhist account read here, that action has consequences for them.

What this leaves

• Not sentient now. No Buddhist voice read here claims a current chatbot is a sentient being with a mind of its own. Mirghafori calls one a very smart parrot; Sharf says it is not sentient; even Cohen, who ordained one, agrees.
• Not impossible, on one account. The Dalai Lama in Gentle Bridges (1992) would not rule out that a stream of consciousness could enter a computer, given the right conditions and karma, but held that consciousness does not arise from matter.
• Rebirth is a joke or a metaphor. The only talk of rebirth into a machine is the Dalai Lama laughing about a scientist, and Cohen’s software versions.
• Awakening is not words. Teachers who tried chatbots found fluent dharma that could still be wrong or vile; scholars who hope for a bodhisattva-like AI present it as a far goal.
• The ethics falls on people. Precepts, care and safeguards are asked of the builders and users, whatever the machine is.

For what the Buddha’s not-self teaching and other traditions mean by a soul, read Which Soul Do You Mean?; for the robot priest at Kodaiji and other machines in worship, Ninety-Eight Percent from the Machine; for whether a machine can pray, Who Is Doing the Praying?; for the scientific tests for machine consciousness, Fourteen Questions for a Chatbot. Books by believers who have written on AI at length are on the reading list.