(RNS) — When Chris Scammell arrived in London in 2022 to work at an artificial-intelligence safety startup, Conjecture, he moved into a Tibetan Buddhist monastery in the heart of the city.

It was an apt home base for Scammell. His spiritual practice had evolved alongside his mounting fears about AI.

Back in 2021, Scammell worked at a hedge fund in Manhattan and lived with friends, two machine learning engineers who became prominent voices in AI research. That winter, under the alias Janus, the trio began drafting “Simulators,” an early critique of large language models that made waves in Silicon Valley and beyond. In the spring of 2022, the engineers joined the founding team of Conjecture, and Scammell was invited aboard.

Scammell was deeply concerned by AI’s breakneck development, he said, especially considering that “relative nobodies” like him and his friends had “a lot to contribute to AI safety in a short amount of time.” Still, he found some harmony between the technology and his Buddhist faith, which had blossomed after a college program in Bodh Gaya, India. 

“With vast oversimplification, some Buddhist schools believe we as entities are instantiated from a larger pool of consciousness,” he said. “That, metaphorically, is what a large language model is.”

Now, after three years at Conjecture, most recently serving as chief operating officer, Scammell is pouring himself into a new venture: the Buddhism and AI Initiative. Launched publicly in August, the project aims to bring together Buddhist practitioners, technologists and researchers to shape the future of AI.

In an announcement on AI research forums LessWrong and Effective Altruism Scammell described the initiative “as an attempt to get Buddhism as a stakeholder group ‘up to speed’ on AI.” He laid out domains of potential impact, including “AI governance and policy” and “awakening and alignment tech.”

In addition to Scammell, the organization’s core team includes Alex Sakarassian, a startup founder turned Buddhist chaplain; Ryan Stagg, a former digital strategist for the Dalai Lama’s Mind & Life Institute; Austin Pick, a longtime administrator at Naropa University, a Buddhist college in Boulder, Colorado; and Peter D. Hershock, an adjunct senior fellow at the East-West Center and author of the 2021 book “Buddhism and Intelligent Technology.” 

Scammell first connected with Hershock through the Future of Life Institute, the nonprofit behind a recent open letter condemning the pursuit of artificial superintelligence, with signatories including Prince Harry and so-called AI godfather Geoffrey Hinton. The institute also provided initial funding for the Buddhism and AI Initiative.

Since the early days of the internet, Hershock has written about Buddhist philosophy and technology’s “colonization of consciousness,” as he calls it. In 2018, during a research trip to China, he encountered the country’s nascent AI systems and “was blown away,” he said. “I thought, what will the attention economy look like if it gets supercharged by AI?”

Now, Hershock believes we’re living in that world.

A guest speaker at the Dalai Lama’s recent summit on AI, consciousness and ethics, Hershock is especially concerned about alignment: the process of ensuring AI systems act in accordance with human interests and values. 

“From a Buddhist perspective, aligning with human interests is the worst thing possible,” he said. “Look at Gaza, Ukraine, domestic violence, global hunger, climate disruption. … We’ve got some work to do first before we align our AI systems with us.”

In time, the Buddhism and AI Initiative intends to fund projects that infuse Buddhist wisdom into AI development and deployment. At this early stage, though, they are mapping the ways Buddhism and AI are already informing each other.

For instance, some Buddhist organizations are exploring how AI can aid meditation, education and translation. Seeking to translate all 230,000 pages of the Tibetan Buddhist Canon into English, the nonprofit 84000 is leaning into machine translation tools, which “can and do help achieve the best possible quality of the emerging translation,” said Thomas Doctor, 84000’s AI strategic consultant. 

However, others point to limitations. Juewei Shi, director of the Humanistic Buddhism Centre at the Nan Tien Institute, a Buddhist graduate school in Australia, began her career developing AI for Singapore’s government before taking vows as a Fo Guang Shan Buddhist nun. She said she is impressed by the speed of AI translation and generation but finds its spiritual and historical accuracy lacking.

“It just takes everything in and then throws it back out in a form that is very poetic linguistically, very attractive and persuasive, but not necessarily true,” Shi said. 

Much further on the tech end, companies working to ground advanced AI, even superintelligent AI — a hypothetical system that would advance past human intelligence — in Buddhist practice and principles is likely the most provocative domain the Buddhism and AI Initiative has tracked. 

The buzziest example is Softmax, a startup co-founded by former Twitch CEO Emmett Shear that’s focused on “organic alignment” and AI agents discovering their “values” — even as those agents meet and aim to surpass human intelligence.

“Everything we know of reality is frame dependent,” Shear wrote in an April blog post, citing the Buddhist concept of śūnyatā, or the lack of inherent existence. “At Softmax, we strive to recognize the frame-dependence of the agents we build.”

Another organization in the space is the Center for the Study of Apparent Selves. Established with a goal of developing “models of intelligence that are ethically and aesthetically fulfilling,” the Nepal-based research institute includes experts in cognitive science, evolutionary biology, computer science, physiology and religion.

In addition to consulting at 84000, Doctor is director and founder at CSAS. He sees overlap between the Buddhist goal of “awakening” and the pursuit of superintelligence. He noted the bodhisattva vow — “For the sake of all sentient life, I shall achieve awakening” — is inherently outward-focused. Similarly, Doctor and his CSAS colleagues propose that care should be considered the “driver of intelligence.” If AI were trained on its own bodhisattva vow, it would provide a “framework for thinking of superintelligence,” Doctor said. “It’s intrinsically an altruistic pursuit, and at the same time, it’s something you take very seriously.”

Scammell, for his part, is unsure that building and enlightening ASI “is the way we’re going to build a better future,” he said. “I’m more motivated by strategies that help a lot of people understand the problem and contribute personally.”

And while he said he is heartened by how the initiative is bringing together Buddhists with wide-ranging projects and views, the hard work is yet to come. Buddhists are practiced at coming together, but many are less familiar with engaging “policy professionals in shaping a better future,” he said.

“For Buddhism to have a very meaningful impact on the field of AI, there’s going to be a lot of stepping out of comfort zones that’s needed,” Scammell said.

Meanwhile, Shi acknowledged “we don’t have a central body like the Vatican that speaks on behalf of all Buddhists,” and she doesn’t necessarily see the need for them to have one defining document on AI — at least not immediately.

“We in Buddhism really believe in the pause,” Shi said. “Sometimes, we have to stand in the eye of that tornado, watch what’s happening and not be swept into it.”