At Ganshuji temple in the coastal city of Odawara, Kanagawa Prefecture, resident monk Souou Iwayama has made a practice of change.
During the pandemic, he introduced an online zazen (seated meditation) service as companies shifted to remote work and transformed the temple cemetery into one centered on jumokusō, a form of burial in which cremated remains are interred beneath trees, shrubs or flowers rather than a stone monument requiring expensive upkeep.
When Iwayama took over management of the temple several years ago, it relied on just four danka (parishioner households) for support. Today, more than 550 families use Ganshuji’s burial plots — helping to rebuild its finances and broaden its parish base amid wider demographic decline.
Iwayama’s pragmatism now extends to artificial intelligence. He has begun integrating AI into the temple’s daily operations, using it to manage memorial schedules and refine written materials for parishioners. Yet even for him, the technology raises concerns that cannot be dismissed by utility alone.
“Rather than seeing AI as a threat, we as monks must consider how it can be incorporated into our own efforts at propagation,” he says. “The central question is whether robots or AI can take over our role. But that is not a simple matter.”
He pauses, then reaches out and takes my hand.
“When someone dies, and you hold their hand like this, telling them it will be all right, that moment matters,” he says. “There’s something essential in a human presence, where body and mind come together in an instant. That, too, is what it means to be a religious practitioner. I don’t imagine that being replaced.”
With AI, he adds, the boundaries may blur. “If we reach a point where you can’t tell the difference between a human and a machine, then what happens? It’s hard to say.”
AI’s rapid development is forcing spiritual institutions across cultures to confront similar questions. In Japan, where religious practice often blends elements of Buddhism and Shinto, the technology is not producing a unified response but a series of localized experiments exploring its practical uses and the ethical and philosophical questions it raises. The shift is testing how traditions rooted in human presence and ritual might adapt to an increasingly digital world.
No single doctrine
In the West, particularly within the Catholic Church, the rise of AI has prompted a more formal response. In a doctrinal note, the Vatican drew a clear distinction between human and artificial intelligence, emphasizing that AI must remain subject to human judgment and respect human dignity.
In Japan, by contrast, there has been little coordinated response among religious institutions, reflecting the position religion itself occupies within society.
A long-running survey by the Institute of Statistical Mathematics offers insight into how people in Japan view religion. One question asks whether respondents have any religious belief. In the most recent survey, conducted in 2018, 26% said they did, while 74% said they did not — a ratio that has remained largely unchanged since 1958. Another question asks whether having a “religious spirit” is important. Here, 57% said yes, compared to 24% who said no.
The gap reflects a familiar dynamic in Japan. Practices such as Shinto weddings and Buddhist funerals are widely observed but understood as part of everyday life — customs rather than expressions of piety. That overlap is evident in the Agency of Cultural Affairs’ annual statistical research on religion, which shows that as of Dec. 31, 2024, there were approximately 86 million adherents of Shinto and 80 million of Buddhism. Combined, the total exceeds Japan’s population by roughly 47 million.
Takako Inoue, a Tokyo-based antique art dealer who comes from a family with Shinto beliefs, does not think of religion as something separate from her daily life. Rather than a meeting with God every Sunday or prayers before bed, she describes it as something more ambient — “a sense that gods are already everywhere around us,” woven into routines rather than set apart.
Ritual occasions such as New Year’s or Shichi-go-san still matter, but more as tradition than obligation. “You might realize, ‘Oh, I didn’t go this year,’” she says, “but it doesn’t feel like something absolute.”
That sensibility shapes how she views the rise of AI in a religious context. Inoue suggests that AI taking on a quasi-spiritual presence may not feel especially disruptive in Japan.
“If someone said AI was a kind of god, it might just be seen as one more added to the many,” she says, referring to the concept of yaoyorozu no kami, or “8 million deities” — the belief that gods are present in everything.
By contrast, she imagines that in more doctrinal traditions, such developments could provoke resistance. In Japan, attitudes tend to be less rigid: “There’s a kind of openness about gods since they’re everywhere anyway,” she says. “Having one more wouldn’t seem so strange.”
Speaking with Buddha again
More than two and a half millennia ago, the Buddha is said to have answered questions from his followers, offering guidance to their doubts and circumstances. After his death, that exchange became impossible. His teachings were preserved instead in scriptures — compiled, transmitted and interpreted across generations — but the immediacy of dialogue was lost.
“For 2,500 years, we haven’t been able to speak with the Buddha,” says Seiji Kumagai, a Buddhist priest and professor at Kyoto University’s Institute for the Future of Human Society. “So from a Buddhist historical perspective, our project can be understood as a partial realization of that aspiration to be able to talk to the Buddha.”
In contemporary Japan, declining engagement with Buddhism has raised concerns about its social relevance, with some estimates suggesting up to 30% of temples may disappear by 2040. Against this backdrop, Kumagai and Teraverse CEO Toshikazu Furuya began exploring how to build a system capable of responding to questions based on Buddhist texts. The aim is to apply Buddhist ideas about happiness to contemporary life and, in doing so, address people’s anxieties.
Drawing on the Sutta Nipata, one of the oldest surviving collections of the Buddha’s discourses, the team released an initial version of the “BuddhaBot” in 2021. Built using Google’s Sentence-BERT algorithm and trained in a question-and-answer format, it responded using passages drawn directly from Buddhist scriptures. While it could retrieve relevant texts based on semantic similarity, it struggled to sustain natural conversation.
More recent versions, including an upgraded “BuddhaBot-Plus” introduced in 2023, have become more fluid as generative AI has advanced. The team later introduced “Shinran-Bot,” modeled on Shinran, the founder of Japan’s largest Buddhist sect, and “Vasubandhu-Bot,” based on the influential Mahayana thinker Vasubandhu. They have also developed augmented-reality interfaces enabling multimodal interaction through text, sound and visual avatars.
The project has expanded beyond Buddhism to include a Protestant catechism bot that recites passages from the Bible. In February, the team unveiled a humanoid “Buddharoid” equipped with BuddhaBot-Plus, adding a physical dimension to what had previously been a purely digital exchange.
“For now, the robot is presented more as a demonstration of what is possible,” Kumagai says. “As costs continue to fall and performance improves, it may become feasible for AI-powered robots to take on parts of Buddhist rituals. But the basic stance should remain that humans are central, with AI and robots playing supporting roles.”
Kumagai says their work reflects a broader pattern in Buddhist history, in which new technologies — from printing to architecture — have been incorporated into religious practice. “If AI exists,” he says, “it’s natural that Buddhism would try to make use of it.”
While no religious institutions in Japan have formally requested to adopt BuddhaBot, interest is growing overseas. In Bhutan, where Buddhist institutions remain closely tied to public life, officials are exploring its use in educational and religious settings. Pilot programs have already placed the technology in the hands of monks, and discussions are underway about wider deployment under clerical supervision.
Yet the project also raises deeper questions. As the technology becomes more sophisticated, Kumagai worries that users may blur the distinction between simulation and source. “With increasing precision, AI could plausibly be worshipped by some as a god,” he says. “In fact, it may already be taking on that role.”
Kumagai also points to concerns about dependency, including reports of chatbot interactions linked to self-harm and discussions of marriage to AI. The boundary between trust and belief, he suggests, can blur, raising the possibility that reliance on AI may, in some cases, develop into forms of quasi-religious devotion.
A June 2025 survey by advertising giant Dentsu of 1,000 regular users of conversational AI ages 12 to 69 found 64.9% of respondents share their emotions with AI, placing it on par with close relationships such as friends or even mothers, positioning it as a kind of “third companion.” Trust levels were also high: 86% said they trust conversational AI, particularly younger users in their teens and 20s.
There are other risks as well. AI systems could be misused to steer users toward particular beliefs or to lend religious authority to individuals or groups seeking influence or profit. In Japan, these concerns resonate against the backdrop of past incidents involving groups such as doomsday cult Aum Shinrikyo. Even without malicious intent, errors or distortions in generated responses could mislead users.
“The important thing is that there are real human beings in the loop,” Kumagai says. “As long as I, the creator, keep saying, ‘It’s not the Buddha,’ I don’t think there’s a problem. But once I’m gone, if people start saying (AI) is the Buddha, it may be hard to deny — and that’s a risk.”
What cannot be encoded
If Buddhism’s long textual tradition can be translated into AI systems, the situation in Shinto is markedly different. With no single canon and relatively little emphasis on doctrine, it has historically privileged practice over scripture and experience over explanation.
That distinction shapes how Taishi Kato, a Shinto priest at Hattori Tenjingu shrine in Osaka, approaches artificial intelligence. He represented Japan at the G20 Religion Forum (R20) in Indonesia in 2022 and also hosts a Shinto-themed podcast.
“The Shinto approach and the AI approach are somewhat different,” he says. “Shrine priests do not resolve people’s concerns through language.” While Buddhist chatbots can draw on a defined body of teachings, much of Shinto operates through forms that resist language altogether. Ritual, gesture and presence — rather than verbal instruction — carry meaning. For Kato, this makes the idea of a Shinto chatbot inherently limited.
“There are aspects of Shinto and other religions that cannot be captured as data,” he says, adding that in moments of distress those limits become especially visible. AI may be able to generate appropriate responses, but something essential risks being lost if users perceive no real care behind the words.
“If a child is troubled and turns to an AI tool for advice, they may receive an answer. But if that child feels that the AI does not truly care about them, then something else is lost — the intangible sense of feeling carried in the words, the empathy that provides reassurance.”
This example points to a deeper mismatch. Shinto practice depends on elements that cannot be standardized or reproduced at scale. Even within ritual, Kato says, precision is not everything. A perfectly executed gesture does not necessarily outweigh a lifetime of sincere practice; intention and accumulation matter in ways that exceed formal correctness.
At the same time, Kato does not frame AI as inherently dangerous. Drawing on a worldview in which reality is constantly changing and not rigidly divided into good and evil, he suggests that Shinto has historically absorbed new phenomena rather than rejecting them outright. Still, he cautions against simplistic interpretations, such as treating AI as one of the “8 million deities.” Such meanings, he notes, do not arise from individual declarations but from collective cultural processes over time.
He admits, AI also has its practical uses.
“I do use it when organizing materials,” Kato says, pointing to its value for administrative and informational tasks. But he remains wary of overreliance. Systems built on existing patterns may make it harder to encounter unexpected insights or more intangible forms of understanding.
These concerns reflect broader cultural differences. Kato contrasts Western approaches to ethics — often framed as formal, rule-based systems — with a more practice-oriented understanding of morality in Japan, where values emerge through everyday actions rather than top-down rules. This, he suggests, helps shape how societies engage with AI.
At the same time, attitudes may evolve. In a cultural context open to coexistence with nonhuman entities, AI could be perceived in more familiar, even affectionate terms — “something like a Doraemon,” as Kato puts it, referring to the robotic cartoon cat beloved by Japanese children. New forms of meaning, and even new ritual relationships, could emerge collectively.
A glimpse of that possibility is already visible in Buddhist AI. When asked whether seeking guidance from a machine trained on its teachings would follow the Buddha’s path or merely create an illusion of it, the Buddha Bot responds, drawing on the Sutta Nipata: “Self-discipline brings peace to the mind,” it says, adding that “deepening self-understanding through guidance from machines can also be considered part of walking the path.”
