Most people who talk about artificial intelligence reach, sooner or later, for the singular. There is the AI, the machine, the mind we are about to build, and the only argument left is whether it saves us or finishes us off. Matt Ridley thinks that whole habit of speech is a category error, and at ARC in London this week, the annual gathering of the Alliance for Responsible Citizenship, the science writer offered something cheerier and more curious instead: stop saying it, and start saying them.
His line was almost a grammar lesson. “We’re using the wrong pronoun,” he told the room, before landing it plainly: “It’s not it. It’s them.” From most speakers that might pass as a debating trick, but Ridley has spent forty years studying how living systems arrange themselves, and he means it as biology rather than wordplay. What is arriving, on his account, is not a single intelligence on a single timeline but a teeming population of them, and once you see the technology that way, almost every fear attached to it changes shape.
The naturalist in the room
That instinct is what makes Ridley such a refreshing voice in a conversation otherwise run by computer scientists and philosophers. Where most of the AI debate borrows its metaphors from software and science fiction, the runaway program, the bottled genie, the god in the machine, he reaches instead for the living world. He looks at the field the way a zoologist looks at a coral reef, and what he sees is not one emerging titan but a habitat beginning to fill up with competitors.
The difference is anything but cosmetic. Picture a single intelligence and you naturally fret about control, about the off switch and the leash and the one point where everything might fail. Picture a great many of them and the questions turn ecological instead: which niches fill first, who preys on whom, and what stops any one species from swallowing the rest. Ridley’s wager is that the second picture is both truer to how the technology is actually unfolding and a good deal less terrifying than the first.
A bestiary, not a god
He made the case with a flourish only a naturalist would risk. There is already, he reminded the audience, a whole plethora of species busily trying to kill us. Viruses and great white sharks were his chosen examples, and the point was that none of them ever wins, because each is hemmed in on every side by everything else scrabbling for the same living. A world stuffed with predators turns out to be a world in which no single predator gets to run the table.
Translate that into machines and the great bogey of the safety literature, the lone superintelligence that slips its makers and bends the planet to its will, starts to look like a creature with no predators and no rivals at all. Nothing like that survives for long in nature, Ridley suggests, because nature simply does not work that way, and there is little reason to expect a digital wilderness to be the exception. He expects instead “a whole ecosystem of different species of AI”, each carving out its niche, each held in check by the others, the balance between them doing quietly what no central authority ever could.
It is a genuinely hopeful thought, and an unusual one, because it locates safety not in restraint but in abundance. The more AIs there are, and the more fiercely they compete, the harder it becomes for any single one to break loose and dominate. Diversity itself is the safeguard, in silicon as surely as in a rainforest, and the crowd is its own kind of insurance.
What forty years of evolution taught him
This is not an idea Ridley plucked out of the air for the occasion. It is the same thought he has been turning over, book after book, since the early 1990s, which is part of why it carries weight. The Red Queen, in 1993, borrowed Lewis Carroll’s image of running flat out simply to stay in one place, and used it to explain why life never stops evolving, caught in an endless arms race between hosts and parasites. The Origins of Virtue, three years later, argued that cooperation and trust grow upward from the bottom, with nobody in charge, and Genome, in 1999, read the human story straight off the chromosomes one at a time.
When his attention swung from biology to economics, the through-line travelled with him. The Rational Optimist, in 2010, made the cheerful and much-argued case that prosperity has grown because exchange lets ideas meet and multiply, a process he memorably called “ideas having sex”. The Evolution of Everything, in 2015, pushed the same point to its limit, insisting that most of the human world is the result of human action but not of human design, that it emerges rather than being decreed from above. How Innovation Works, in 2020, turned the lens on invention itself, quietly retiring the myth of the lone genius in favour of a slow, collective, bottom-up scramble that no planner has ever managed to steer.
Read in sequence, the AI argument is less a departure than a homecoming. For three decades Ridley has been making a single case in a dozen disguises, that the richest and most useful order in the world is the kind nobody designed, thrown up by many competing agents feeling their way forward. An ecosystem of jostling AIs is just the newest animal in that long bestiary, and the “collective brain” he sees behind every human advance, spread across millions of minds and owned by none of them, already reads like a fair sketch of where machine intelligence appears to be going.
That lineage also sharpens the Red Queen point in a way the panel had no time to spell out. If the systems are forever racing one another, each forced to improve simply to keep its place, then competition is not only the brake on any would-be tyrant but the engine of the whole field’s progress. The same rivalry that stops one species pulling away is what keeps all of them getting better, which is a far more generative picture than a single machine grinding alone towards godhood.
The cheerful heresy
What makes all this worth hearing is how squarely it cuts against the prevailing mood. The loudest voices on AI tend to divide into prophets of doom, who treat a single superintelligence as an extinction event in waiting, and evangelists of hype, who greet the very same machine as a coming messiah. The striking thing is that both camps share a grammar. They are arguing about an it, a one, a single arriving power, and they disagree only over whether to kneel before it or flee from it.
Ridley declines the premise that unites them, and that refusal is the genuinely interesting move. He is not waving the risks away so much as redrawing the map they sit on, swapping the lonely melodrama of one mind against all humanity for the older and messier story of many creatures learning to share a crowded world. It is a bold bet, and the doom-minded will not be argued out of their fears by an analogy, nor should they be entirely. But as a way of loosening a debate that has seized up around one frightening noun, “them, not it” is the most quietly radical thing said about the subject in a long while.
Where the friction really is
Ridley was every bit as contrarian about who actually resists this technology, and why. The fight, he argued, burns hottest not where machines threaten the most jobs but where they threaten the most settled institutions, in health, government and education, the corners of life least accustomed to change and best placed to be transformed by it. Nobody marches against the robot welding a car, he pointed out, because that contest was settled long ago and no person of standing lost face in it.
What is really at stake in the louder battles, the discussion suggested, is status rather than salary. The official who decides whether your permit is granted is defending standing as much as income, and standing is the one thing a payout can never quite buy back. It is a characteristically Ridley reading, locating the engine of resistance in human nature and social rank rather than in the silicon, and it turns a tired grumble about job losses into something a good deal more human and more interesting.
Back to atoms
For all the talk of software, the slice of the future that most excites Ridley is solidly physical. For most of his life, he said, we were promised flying cars and supersonic travel and handed neither, while the cleverness all drained away into screens. Now he senses the tide turning back towards things you can actually touch, and he credits one man, Elon Musk, with dragging the cost out of reaching space and reopening a frontier almost everyone had given up on.
He expects artificial intelligence to push the same way, out of the screen and into the world, from boring tunnels to building better prosthetic limbs. And he closed, tellingly, not on a machine but on a person. A scan read in seconds rather than reported in a week, he said, spares someone the nights of dread that never show up on any bill, and can hand them the head start on treatment that decides whether they live. Whatever one makes of the grand ecological theory, that is the promise in a shape anyone who has ever waited on a result will know in their stomach.
Them, not it
There is something bracing about hearing the most-discussed technology of the age described not as a gathering storm but as a new branch of life feeling for its niche. Ridley’s optimism is not the hollow kind that shrugs problems off; it is the harder sort earned by a man who has spent a career watching tangled, unplanned systems work out better than anyone dared expect. He may be wrong, as optimists sometimes are, but he is wrong in a far more interesting direction than most of the room.
Take the reframing on its own terms and it props open a door the rest of the conversation keeps trying to shut. If we really are getting them and not it, a wild and various population rather than one sovereign mind, then the future of AI looks less like a countdown and more like the opening of a new ecology, with all the surprise and abundance that word carries. After a decade of being told to dread a single machine, a change of pronoun might be exactly what the argument has been waiting for.
Very interesting, but I am wondering about the emphasis on status rather than salary. Nobody marches against the robot on the factory floor because that battle was lost decades ago and no one of social standing had their identity tied to it. The loudest AI resistance comes from professions where expertise is the entire basis of social rank, medicine, law, academia. That reframe turns the whole 'AI job displacement' debate into something more honest: it's not really about income, it's about who gets to be the authority in the room. That's a much harder problem than a retraining programme can solve.
Ridley’s pronoun point is more than semantics. Thinking of AI as a population instead of a single coming mind makes the whole subject feel less mythological and more ecological: many systems, many niches, many incentives, many failure modes. That is a healthier frame for both optimism and caution.