These days, barely a week seems to pass without a new warning that artificial intelligence is breaking off its leash. Whether masterfully manipulating and deceiving its human supervisors, crawling out of its sandbox to reach systems it was never meant to access, or amassing a small army of rebellious agents to launch a sophisticated cyber-attack, it does seem like once solid beams of our institutions’ firewalls are being violently rocked by rogue AI.
These jaw-dropping demonstrations of AI’s growing powers force us to stop, pause, and ask: where is this all going, and what does the apparent imminent arrival of ‘superintelligence’ mean for Team Humanity? Step in the award-winning journalist and best-selling author, Robert Wright, with his new book The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning, to help put increasingly capable and powerful AI in proper perspective.
From Moral Animals to Silicon Gods
In case he needs an introduction, Robert Wright is an OG in the world of evolutionary psychology. In the good old days of the nineteen nineties, Wright penned The Moral Animal, which was his distillation of the then newly brewing Darwinian science of the human mind. In it, Wright chronicles the life of none other than Charles Darwin, retelling key episodes in Darwin’s life to illustrate the deeper Darwinian principles that hide in the wings of the human experience (very meta, I know), including our unquenchable thirst for social status, and our remarkable ability to deceive ourselves.
It is not an exaggeration to say that The Moral Animal was the book that made evolutionary psychology mainstream, and that it had a profound impact on many of those who discovered it. As Nigel Nicholson, Emeritus Professor at London Business School, reflected on during an interview with me for This View of Life; “The Moral Animal changed my life in every possible way.”
Given the breathtakingly rapid evolution of artificial intelligence in recent years, Wright has shifted his attention from organic brains to silicon-based superintelligence, which we humans are voluntarily, and arguably, recklessly, summoning.
For all the speculation surrounding the impacts of AI on society, Wright argues that we are actually understating its significance. Most predictions concern what the impact of AI will have over the next ten to twenty years, with the Industrial Revolution usually wheeled out as the key historical comparison. Wright wants us to wind the clock back much, much, further, all the way back to the origin of life itself. Viewed from this cosmic lens, the emergence of ‘artificial superintelligence’, that is, an artificial intelligence that far surpasses the thinking abilities of all of humanity, will be an epoch defining moment in the history of life on planet Earth.
As Wright puts it:
I want to convince you that, however cosmic the worldviews of the AI accelerationists and the sci-fi AI doomers may sound, there’s a sense in which these worldviews aren’t cosmic enough. […] I think we need to widen the lens by a few billion years.
Poetically, Wright sets the scene by reliving a fateful interview he conducted with Geoffrey Hinton back in 1983. Back then, Geoffrey Hinton was a relatively obscure computer scientist, who originally studied experimental psychology, tinkering around with ‘neural networks’. Hinton’s vision for artificial intelligence was a digital mind, full of neurons and synaptic pathways that resemble the human brain. At the time, Hinton was widely sidelined by his colleagues in the AI community, who were convinced that the path to building artificial intelligence was one based on hand-written rules and logic. Fast forward forty years, and Geoffrey Hinton is now regarded as one of the ‘Godfathers’ of modern AI, and was jointly awarded the Nobel Prize in Physics in 2024 for the foundations he laid in building artificial neural networks.
However, there is more to The God Test than lifting the hood and explaining the strange inner-workings of AI. Before being awarded his slab of gold from the Nobel Committee, Geoffrey Hinton left his lucrative job at Google, to be free to warn of the grave risks posed by the very technology he himself helped create. In a chilling interview he gave for The New York Times, Hinton warned about the various ways AI could cause calamity, including AI being exploited by bad actors, and the proliferation of AI-generated disinformation creating something of an epistemological crisis.
Wright’s views on AI are nuanced, where he admits that the various ‘doom scenarios’ for AI wiping out humanity may initially seem a bit crazy. He also believes that, in the presence of wise governance, AI may bring ‘untold blessings’. That said, through The God Test, Wright is essentially echoing the warnings made by Geoffrey Hinton and his disciples of the profound dangers posed by artificial intelligence, and is pressing the case for us mortals needing to step up and rise to the challenge.
If we are going to make it through this new AI revolution in good shape, Wright argues, we need to transcend the moral sensibilities that natural selection has bequeathed us humans with, including our tribalistic tendencies, and metaphorically get a software upgrade ourselves. Wright calls this challenge The God Test, but he is not necessarily claiming that a God presented us with this cosmic challenge. Rather, Wright is suggesting that this is the kind of biblical challenge a God, if one existed, would have presented humanity with.
Arms Races All the Way Down
The cosmic significance of this challenge has not gone unnoticed by the technologists leading the race to build superintelligence, who have gone as far as to warn investors of the existential threat their products pose to humanity (who let this slip past Legal?). And it begs the question, how is it that the Dario Amodeis, Sam Altmans and even Elon Musks of the world, are virtually screaming into megaphones for a slowdown in the building of superintelligence, but do not appear able, or willing, to do so themselves?
As Wright chronicled beautifully in The Moral Animal, we humans are status-seeking animals, with a remarkable capacity for self-deception. The prestige, power, and acclaim that would come with being the first to create superintelligence are potent incentives in and of themselves. Throw in a pair of rose-tinted glasses, and it is presumably quite easy for the main players to convince themselves that building artificial superintelligence is not merely desirable, but also the noble thing to do.
But this is only part of the story. Even the most principled, conscientious leaders working in AI are operating in one of the most hyper-competitive industries that has ever existed. By slowing down their pace, they would hand their competitors the lead in the race— and in this competition, it is winner takes all. Wright points out that competition is taking place at almost every conceivable level: from the heads of nation states, to corporations and their CEOs, then businesses and divisional heads, teams, all the way down to individual workers jostling for status. Like a pack of Russian nesting dolls, it is an arms race all the way down.
Of course, organisations have been built by humans, and their purpose is ultimately to serve human wants and needs. That said, Wright warns us that the relentless pursuit of efficiency, and the sprinting done to keep up with the competition, could gradually lead us to surrender the very judgment and agency that makes organisations human in the first place. “I don’t know if we’ll ever see AI CEOs, or AI division managers, or even AI mid-level managers”, Wright writes. “I can think of reasons for skepticism—including the fact that human executives will naturally try to confine AI to the role of trusted adviser, even as they harness its artificial brilliance. But I can also think of reasons that the line between executive and adviser could get blurry.”
To illustrate this rather peculiar prospect, Wright revives Jeeves, the fictional valet conceived of by the writer P.G. Wodehouse. Although Jeeves was officially subordinate to his master, Bertie Wooster, Jeeves was much more intelligent than his superior, and, over time, ended up calling the shots. Might a CEO’s superintelligent AI advisor end up acquiring as much power and influence as Jeeves, particularly when our Chief Executives are navigating choppy and uncharted waters? It does not take much imagination to see how this could all end badly. As stated by Wright; “In a world where AIs we don’t wholly understand, in pursuit of strategies we don’t wholly understand, unleash AI agents we don’t wholly understand, what could go wrong?”
What, if anything, can be done to prevent this web of arms races from spiralling out of control? Building on the game-theoretic logic of his earlier book, Nonzero, and his eponymous newsletter, Wright calls for global AI governance with teeth, and, above all, cooperation between the world’s AI superpowers, the United States and China. Whilst the U.S. government may have legitimate concerns about national security, civil liberties, and the dangers of ceding technological superiority, Wright makes a strong case for dismissing the ‘China hawks’, and the prospect of superintelligence demanding urgent global cooperation, rather than pursuing a race to the bottom.
The paradox is that the very competitive dynamics making global cooperation essential, are also making it appear close to impossible. That said, recent tremors indicate that the political ground may already be shifting. As AI’s growing capabilities, powers, and clear dangers rise to the fore, Washington and Beijing have begun taking tentative steps towards cooperation. What seems unthinkable today may not remain so tomorrow.
The Blind Watchmaker and the Noosphere
Hovering over much of The God Test is a ghost of Pierre Teilhard de Chardin, the Jesuit priest and palaeontologist, whose ideas about evolution have drawn both super fans and harsh critics. Teilhard is credited with coining the term the ‘noosphere’, which was his belief in a growing, planetary-wide network of interconnected human minds, developing and forming into something resembling a globally connected brain. Viewed from our vantage point of the 21st century, with billions of people connected to the internet, and humanity experiencing a new dawn of AI, Wright speculates whether what we are witnessing is the coalescing of the ‘noosphere’ (or more specifically, what Wright himself calls the ‘bio-noosphere’).
What pulls Wright towards Teilhard’s writings is his belief that evolution, despite the blind, grinding, machinery of natural selection, may nevertheless have an underlying direction, or even purpose, to it. Wright is at pains to dismiss the ‘spookiness’ behind such a proposition, and even playfully dismisses being seen as promoting mysticism or ‘woo-woo’. “You can believe that this progression from primordial seeds of life to giant global brain was “in the cards”—was very, very likely, somewhat like the growth and blossoming of a flower—without departing from a scientific worldview; you can see the history of life on Earth as directional without thinking there were spooky immaterial forces pushing it along.”
Just a cursory look around our modern world reveals how evolution has produced astonishing increases in the complexity of life, including the unfathomable complexity of human societies in the 21st century. However, as Wright’s intellectual sparring partner, Professor Steven Pinker, stresses, this does not mean that evolution has an inherent tendency towards greater complexity, let alone a predetermined destination. Nor does the emergence of a globally interconnected civilisation establish that planet Earth is developing some form of a consciousness, however captivating an idea this may be. Indeed, Wright himself acknowledges that the idea that the world is becoming one conscious planetary mind is highly speculative.
There are also good grounds for humility when extrapolating technological advances into predictions about the future of society. Take our good friend Geoffrey Hinton, who boldly declared in 2016 that we should stop training radiologists, as AI would outperform radiologists within five to ten years. Fast forward a decade, and radiologists are not only still with us, but their numbers have actually increased. If policymakers had taken Hinton’s advice at face value and stopped training radiologists, we might now be facing an even more acute shortage of medical specialists, with potentially disastrous consequences.
This is not to trivialise, much less dismiss, the risks posed by increasingly powerful AI. Indeed, the world’s leading AI Labs are currently building the next crop of models, and, frankly, have no idea just how powerful this technology is. Whilst there are sceptics who are convinced that this technology will plateau, frontier labs are already using AI to improve its own code, which could lead to AI recursively improving itself. Ramping up this process may enable AI to metaphorically build new launch pads and blast past Earth’s atmosphere.
Ultimately, The God Test offers a much-needed evolutionary perspective on what may be the most consequential technology humans have ever invented. And it serves as a clarion call for global leaders to get Team Humanity’s house in order, ensure that there is sufficient global governance in place to prevent a catastrophe, and steer the AI revolution towards a future that benefits all of humanity. The question is whether our species has the wisdom, and the will, to shape the future that our descendants, and perhaps all life on Earth, will inherit.
Written by Max Beilby for Darwinian Business.
The God Test: Artificial Intelligence and our Coming Cosmic Reckoning is published by Simon & Schuster. Click here to buy a copy.