Robert Wright’s Princeton library has seen many remote visitors. The veteran journalist and author — formerly at outlets including The New Republic, Time and The Atlantic, with five books published and a sixth out this June — has been holding court in front of his packed bookshelves for decades.
A writer liberated by the blogging revolution, he founded the video-interview platform Bloggingheads.tv years before podcasts became the new daytime television, and he has also gone his own way for decades. This editor was a devoted watcher, decades ago, when a young who’s who of literary types Skyped in to talk with Bob, from Ezra Klein to Ta-Nehisi Coates and Megan McArdle. So it was a touch surreal to see those familiar bookshelves and hear that familiar voice corresponding with Fortune. When I told him as much, he responded with his familiar self-deprecation: “Well, your ship has come in.”
For the last five years, Wright has made his home at the widely read NonZero newsletter and podcast, building on a 26-year-old book of the same name that argued human history is shaped largely by non-zero-sum dynamics—by encounters in which one side’s win isn’t necessarily the other side’s loss. The situation is usually nonzero, in other words. It’s informed by Wright’s writings on Buddhism throughout the years, which he sees less as a religion and more as a way of thinking.
That is exactly why, he told Fortune during a recent interview, the current situation is so alarming. That situation is not just artificial intelligence, according to Wright—but the lost concept of “enlightenment” in the 2020s.
“When you look at all the fronts — economic, cultural, political, personal, family, psychological and so on — it is going to be an earthquake,” Wright said. “That’s my prediction. And we better be ready for it.”
A belated ‘oh sh-t’ moment
In 1983, Wright was a young journalist on assignment, sitting down with Geoffrey Hinton, years before the computer scientist was dubbed the “godfather of AI” and years before Hinton resigned from Google DeepMind expressing regret at his creation. As Wright explains in his forthcoming book, The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning, he completely missed the significance of what he was being told about this concept called “neural networks.”
“I just did not — I was not remotely close to getting the picture,” Wright said. He didn’t know at the time that this computer tinkering would lead to what he now considers an inflection point — not just economic but civilizational. The grand test ahead of us now, he told me, is whether we are qualified to play God, and he’s not sure that we’ve shown we’re capable.
Wright’s reckoning with AI’s true power came not in 1983 but in 2023, when he began experimenting with ChatGPT-4 while under contract for a different book entirely, one on cognitive empathy. One experience made him think of Hinton: he gave GPT-4 an elaborate, layered scenario about a student and a professor, asking what the student would feel at the end. The model’s one-word answer: schadenfreude.
“I thought, ‘oh sh-t,’” Wright recalled. “It really understands human nature and is capable of putting itself in the perspective of another human being.”
The experience sent Wright back to a 2018 lecture Hinton gave on how large language models actually work. The neural network that Hinton pioneered “will kind of reverse engineer the human mind,” Wright said, cautioning that he hasn’t heard many AI researchers put it in quite those words. That includes certain cognitive functions that natural selection engineered over millions of years, he added with astonishment.
“That was my big misunderstanding back in 1983,” he added. “I was assuming that if it was going to get good at handling language, we would have to implant in the machines the human understanding of the connection between the words and the meaning.” But we didn’t have to do any of that. “It just kind of quote ‘figured out’ that if it’s going to get good at predicting how sentences will end or predicting anything else … it is going to have to have a way of representing the meaning of words.” And he offered up a compelling example: it helped figure out an ambiguous MRI report on his own cancer.
The doctor and the machine
Wright is currently cancer-free, he told me. His throat cancer had spread to his lymph nodes, but after successful treatment it hasn’t come back since a surgery in 2025.
At one point, he told Fortune, reiterating a story he related on NonZero in July 2025, he was trying to interpret a confusing doctor’s report, so he fed it into various chatbots. They said it looked like something was worth investigating.
He pressed further, flagging a strange sentence, and Anthropic’s Claude caught what his radiologist had missed. “Claude said, ‘Well, the radiologist meant to put the word ‘no’ at the beginning of that sentence: ‘no abnormalities found.’”
AI caught a crucial grammatical error—a human error—that could have scrambled his understanding of his recovery.
“They say AIs hallucinate, they make mistakes,” Wright said. “Humans make mistakes. This human made a mistake and he’s a doctor, okay, and about a pretty big thing.” The question isn’t whether AI will “take our jobs and become our lovers and do everything else,” as Wright put it: it’s more about whether AI is “in any sense better overall than the people that were filling these roles.” The answer to that is so obvious that Wright doesn’t even have to say it.
“The thing doesn’t have to be perfect by any means to replace a human,” Wright said, “especially given the fact that it’s going to be a lot cheaper and faster.”
Wright’s thinking on AI is similar to his stance on Buddhism: not religious, but a secular, Westernized variant based on journalistic investigation and first-principles thinking. His 2017 book Why Buddhism Is True describes his engagement with Buddhism as a long personal journey predating the book itself, rooted in his earlier work on evolutionary psychology. Wright’s explicit aim is not to bring readers into Buddhism as a religion, but to argue that its core psychological insights — particularly around suffering and perception — are validated by evolutionary biology and modern science.
The ‘suicidal ideology’
If Wright’s earthquake thesis puts him broadly in alignment with figures like Anthropic CEO Dario Amodei and Microsoft’s Mustafa Suleyman — both of whom have warned of dramatic near-term disruption — his sharpest departure from Silicon Valley’s AI establishment is on geopolitics.
Amodei, in particular, has framed AI as an existential race with China, arguing that American dominance is essential to ensuring the technology develops safely. Wright called this a “suicidal ideology.”
“Dario sees us as in this existential AI race with China, and I personally think that mentality will lead us to a very, very, very bad outcome,” Wright said.
He believes Amodei is “sincere, more than most of these people,” but argued the entire framing is dangerously wrong. For Wright, true to his Nonzero beliefs, the only path to managing AI safely runs through global coordination, not competition. An arms race logic, in his view, forecloses exactly the kind of cooperative governance that could prevent catastrophe.
It is, he acknowledged with a wry smile, a very un-Buddhist way to approach an existential problem: each nation clinging to its own perspective, certain of its own righteousness, blind to the view from anywhere else.
He said his frustration with Anthropic sharpened recently following the company’s public acknowledgment that recursive self-improvement — AI systems autonomously improving themselves — may be approaching. The company gestured toward the need for global coordination but offered no concrete plan.
“Wait a second, Dario: you above all have been predicting we get to exactly this moment for years and you guys don’t have a plan?” Wright asked, peppering in a rather colorful adjective. “You haven’t devoted like a hundred dollars to looking into what it would take?”
He noted that Anthropic is due to IPO soon at a nearly trillion-dollar valuation. “Isn’t that because the financial incentives point in the other direction?”
Anthropic declined to comment.
A God test, not a doom prediction
For all his alarm, Wright resists being categorized as a pure doomsayer, and his trademark self-deprecation and good humor are on display throughout our conversation.
The title The God Test is partly a biblical metaphor, he said: throughout the Old Testament, salvation is offered conditionally. Shape up, or face consequences. The choice is real.
“We face an interesting and climactic moment, in a certain sense, in the whole history of evolution,” Wright said. “We’re creating a whole new form of intelligence — hasn’t happened before — in a new substrate that’s not carbon-based. And if you wanted to bring good things and not overwhelmingly bad things, you better recognize immediately that you’re proceeding too fast and recklessly, and you’re gonna have to get together as a cohesive global community.”
He pointed to two recent developments as cautious grounds for hope: a renewed U.S.-China dialogue on AI safety, and the Trump administration’s reversal on reviewing powerful AI models — a position it had previously mocked under Biden. “Change can happen,” Wright said. “I just hope the progress continues to come from developments that don’t get tons of people killed.”
The only way to pass the God test, he added, is to model what he calls “enlightenment.” He doesn’t mean it in the 18th-century, Voltaire and salons and wigs sort of Western sense, he clarifies, but something closer to the Buddhist conception. “It really amounts to just clarity of perception and thought,” he said. “Transcending all of the kind of cognitive and perceptual biases that are built into us — that get in the way of a balanced view of the world. Like the world is viewed from some perspective other than yours.”
Wright brought up “the view from Mars” on the human condition and what that must look like, which he admitted is hard to hear without thinking of Elon Musk — whose SpaceX is the first company with the explicit mission of making David Bowie’s vision a reality: life on Mars.
“I mean, Elon Musk seems to me the opposite of enlightened,” Wright said. “It’s worrisome that the world’s richest man is so incapable of getting outside of his own perspective.”
The salvation for this unenlightened age that Wright sees the world living through, he added, will come from “lots of people, including influential people, getting better at viewing the world from outside of their own perspective.”
This editor asked Wright if he was arguing for something like a Star Trek future, a federation of planets under a one-government system, managing problems peaceably and resolving conflict. He pointed out that Star Trek can’t be seen outside of the context of its 1960s premiere occurring just 20 years after the catastrophe of World War II. “That’s the thing, catastrophes lead to dramatic and ambitious thinking about restructuring orders … I hope it doesn’t take a true catastrophe in this case to get us to take international coordination seriously.”
He said he reaches for another science-fiction classic on the AI discussion: The Day The Earth Stood Still. Like Star Trek, it was a nuclear deterrence drama but it involved aliens visiting with a message for Earth to “get your sh-t together and form a global community and get things under control.”
In a way, he said, AI is sending a message “maybe not quite that dramatic but along the same lines.” He returned to the Bible metaphor, full of instances where salvation is possible, “but you have to get your act together, you have to become better people … you’re going to have to upgrade your game morally and in some sense spiritually or bad things, very bad things will happen. But very good things can happen, if you do the right thing.”
