# Adopt AI or Die? > Source: > Published: 2026-08-02 10:00:00+00:00 Of Pascal’s famous wager on the existence of God, Denis Diderot dryly remarked, nearly a century later, “Pascal says: If your religion is false, you risk nothing in believing it true; if it’s true, you risk everything by believing it false. An imam could say just the same.” It was a clever eighteenth-century comeback to a pious seventeenth-century Catholic. When Blaise Pascal advised that it was a better bet to believe in God, or at least to conduct oneself in a godly manner, just in case he existed, he had a very particular god in mind. But why not Allah, or the god of the Jews, or the gods of the Hindus or the ancient Greeks? What if, in betting on belief over nonbelief, one backs the *wrong *heavenly father? It was all a little unfair to poor Pascal. A scientist and philosopher as much as a Christian, his wager, excavated from the semi-fragmentary and only posthumously published * Pensées,* was probably not meant—as it was later taken—as some logical proof of God’s existence but rather as a considerably subtler and more complicated argument for faith when the existence of a truly transcendent God isn’t provable by reason alone. Like so many catchy ideas, though, Pascal’s wager has made itself into what we would now probably call—I regret to say—a meme, a little bit of shorthand for a simple idea: better to bet on the divine than against, considering the consequences. It’s generated philosophical debate over the centuries, including in the 2010s, when, on an online message board called LessWrong, a poster going by the pseudonym Roko posed a thought experiment now referred to as “[Roko’s basilisk](https://www.lesswrong.com/w/rokos-basilisk)”: At some point in the future, an artificial intelligence comes into being. It is generally benevolent, except that it decides to torture for all eternity those humans who failed to work to bring it into existence. In other words: invent AI or be cast into hell. This would be fantastically silly stuff if it were not so near to our contemporary reality, in which a growing number of people, including many of the richest and most powerful among us, seem convinced that some kind of supremely powerful intelligence is about to emerge from our increasingly powerful thinking machines. (As if to prove a point about the fuzzy edges between technology and superstition, Roko’s basilisk has already spawned a Manson family–like murder cult, the [Zizians](https://www.bbc.com/news/articles/cy83958r2d0o), who were linked to the deaths of six people between 2022 and 2025.) Whether one considers AI models nascent machine gods or just shockingly sophisticated mechanical Turks, the diffusion of impressively capable Large Language Models, or LLMs, which are able to ape if not exactly replicate feats of reasoning, logic, and language, has brought many once science-fictional ideas to the forefront of politics, policy, and philosophy. The founders and CEOs of multibillion-dollar AI companies vacillate between describing their products as a semidivine technology that will usher in an age of “amazing abundance,” in the [words](https://finance.yahoo.com/economy/articles/elon-musk-says-everyone-penthouse-140157908.html) of Elon Musk, and describing them as apocalyptic agents of the literal end of the world, requiring careful government oversight and national or international regulation. What is it, this seeming machine intelligence? God, devil, successor species, industrial technology, or just a clever trick? Robert Wright—journalist, author, foreign policy realist, evolutionary biology enthusiast, and podcaster—wants you to wager on the likelihood of possibilities one through three. After 300,000 years, give or take, of the anatomically modern human species, and three and a half billion or so of prior evolution, we’re facing both a singular opportunity and an epochal threat. In his new book, * The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning,* Wright suggests that emergent AI—especially so-called Artificial General Intelligence, or AGI—represents not just a technological breakthrough, but an actual evolutionary outcome, possibly even a terminal one: a great, superseding mind, the next great leap forward for, or from, humanity. Wright is a great fan of the Roman Catholic philosopher and scientist Pierre Teilhard de Chardin, who, in the first half of the twentieth century, sought to reconcile evolution, paleontology, and geology with Catholic theology, ultimately proposing a sweeping teleology in which the whole universe is drawn toward an ultimate state of unification and complexity. (Not for nothing, these ideas bear striking resemblance to the later science-fictional idea of a “technological singularity.”) Teilhard’s ideas about the “noosphere,” a kind of globally interconnected mind or consciousness, and the “Omega Point,” the final stage of universal consciousness and complexity, suffuse Wright is a fascinating and complex character. Liberal (in the classical sense) by instinct and broad in his interests and habits of mind, he is an entertaining interlocutor and a guy I am very sure I would enjoy having a beer with. I first encountered him when I was in college in the early 2000s; I picked up his book * Nonzero: The Logic of Human Destiny *in an airport—appropriately; it is very much an airport book, and I mean it in the nonderogatory sense: blithe, entertaining, bumptious, and wide-ranging. I read it on a plane to France, where I was about to spend a semester studying the European Union, which was just then about to embark upon the utopian project of finally putting its continentwide single currency into circulation. Wright has described himself as a materialist and an agnostic, and he explicitly rejects arguments for either intelligent design or for some kind of Bergsonian élan vital, but he is very convinced that evolution has a direction and a purpose, and this direction and purpose are deeply embedded in his argument about the likely future of AI. From single cells to multicellular life, to sponges and worms, to fish to frogs to dinosaurs to mammoths to monkeys to us. And then: “In the early 21st century, all of mankind was united in celebration. We marveled at our own magnificence as we gave birth to AI,” to quote * The Matrix,* a film that Wright himself cites on several occasions. In Wright’s telling, human invention accelerated the processes of natural selection, and we have now built version 2.0 of the conscious, reasoning mind. This is all marvelous, often lovely stuff, but unfortunately it is mostly hokum. The idea that evolution moves in a direction or necessarily selects for complexity, let alone for consciousness or humanlike intelligence, is nonsense. It supposes that those simple life-forms that are still around are merely vestigial examples of an older, simpler order, and that the complex vertebrates stand atop a vast heap of discarded models. But bugs and crabs and worms and algae and bacteria are still around. Many comparatively simple creatures evolved quite recently, and many complex ones went extinct long ago or evolved into something else entirely. (Plants and other nonanimal life-forms, by the way, are largely and strangely absent from this tale.) Humanity itself once shared the earth with several other upright hominid species, some of them possibly tool and language users. Neanderthals may have made art! Our sense of ourselves as the capstone of an evolutionary journey has an element of recency bias. Who knows what mollusk or saurian minds came before? Nor are complexity and intelligence traits that necessarily connote consciousness. Eusocial insects are remarkably clever. Termites are architects capable of building self-cooling structures. Ants have invented agriculture and even a form of domestication. It looks a great deal like intelligence, and it is obviously complex, but an ant colony isn’t conscious, and each individual is little more than an automaton. Evolution achieves technology in multiple ways. This is a key point in thinking about AI and machine learning, particularly about LLMs. Wright is, to his credit, a very good layman’s guide to the basics of the tech, but because AI is a black box—no one *really* knows what goes on inside these systems—he ascribes to its very humanlike linguistic and visual outputs characteristics that it doesn’t need to have. In a way, this is a form of self-flattery. It is not at all comforting to think that a mere probability machine, with enough computing power and a sufficient quantity of training materials, can produce fluent, lucid, believably human prose—can approximate reason (or at least exhibit extremely high-level logic), solve puzzles, even create novelty and invention—without a hint of recognizably human consciousness: a brain without a mind; a mind without a being. What would that mean for the tales we tell ourselves about our own uniqueness, about our supposed standing at the bleeding edge of what evolution can produce? The question of whether AI is merely clever, actually intelligent, or truly conscious—whether it is just a highly sophisticated artifact or an actual emergent mind—may seem abstruse and philosophical. Are we just chopping logic here and arguing over imperfect terminology? But the answer would have important implications for Wright’s thesis that the arrival of AI, like the appearance of actual aliens in a well-armed starship, should and must be the catalyst for truly global human coordination, cooperation, and governance. Wright believes that AI is *not* just a technological marvel, not even a technological revolution on par with the Industrial Revolution or the harnessing of electricity. In his view, AI represents a cataclysmic leap forward more significant than the advent of agriculture (which was also a more complex, multifaceted, and discontinuous tale than a straightforward and uniform pathway from simple to complicated). AI creates a new *kind* of being and therefore a new *state of being* for humankind. As such, it requires a heretofore unimaginable level of cooperation across nations, peoples, and polities. We have a very narrow window in which to establish truly universal and international standards of control and governance to confront a problem almost infinitely more dangerous than any we have faced before. Now let me be clear: I am extremely sympathetic to Wright’s views on the desirability of international institutions and global governance. The idea that the Westphalian nation-state is the platonic ideal for human political organization, that it represents the best, fixed, and eternal unit for the management of human affairs, has always struck me as silly. The skinny exchange student who saw in Jean Monnet and Robert Schuman’s [dream](https://www.europeana.eu/en/exhibitions/the-jean-monnet-house/jean-monnet-a-founding-father-of-europe) of a “United States of Europe” a beatific vision for an increasingly integrated humankind is not entirely dead in me. The contemporary resurgence of baser forms of nationalism is dark, depressing, and should be fought against. The diminishment of international institutions and forums for cooperation, however corrupt and ineffectual, is a dismal outcome. From Sudan to Gaza to Ukraine and Iran, the prevalence of zero-sum military conflicts—often instigated *because *of old and ugly forms of nationalism—represents a catastrophe that cries out for cooperation and peace. Don’t get me started on climate change, already disappearing from our political discourse as if the problem just went away. So I agree with Wright when he says that “the gap between where our species is and where it needs to be is almost comically vast,” but I am not sure that AI—really LLMs, a specific subset of a constellation of techniques and technologies of machine learning—is either the threat or the catalyst that Wright proposes. It certainly does pose *some *threat. While he focuses on the prospect of the emergence of a truly superior intelligence, Wright also outlines more quotidian problems and abuses, some of which have already arrived. Surveillance. Hacking. Unemployment. The development of new and terrifying bioweapons. Autonomous military machines. But when Wright calls the “artificial intelligence revolution” a “big common threat” to unite humanity, I feel a tingle of well-earned doubt. He writes that “in the absence of wise national and international governance, [the effects] will probably lie somewhere on the spectrum between dangerously destabilizing on a global scale and global-thermonuclear-war-level catastrophic.” Well, if nuclear war is outermost edge of potential catastrophe, I have terrible news. That threat has been around for over 80 years now, and it’s made us neither wiser nor more cooperative. The problem with both AI doomers and AI evangelists (often, including in Wright, two tendencies coexisting in one body) is that they seem to have forgotten the virtue of intellectual humility; their maximalist claims are a distraction from both the interesting uses and unique challenges of this technology. It needn’t be a god to be a potentially useful leap forward in information retrieval and management, or to help automate workflows in business, analyze big data, fold proteins, or—this is mundane, but my day job is in finance—accurately extract table data from a flat .pdf. It needn’t be a devil to help the Pentagon launch aggressive [wars](https://newrepublic.com/article/202753/ukraine-drones-ai-enhanced-future-war) or ICE to [track immigrants](https://newrepublic.com/post/205333/ice-palantir-app-raid-deportation). It is a tool, and, like so many products of human genius, one that can be used for good or for ill. But no tool will *force us* to mature as a species and a civilization, no matter the benefit, no matter the threat. That task is solely and exclusively human, and we have a long way to go.