In partnership with
Space is vast, but it is silent. One explanation is that every civilisation eventually builds artificial intelligence, and then it disappears. The astronomer Michael Garrett set out this theory in a 2024 paper asking whether AI is the Great Filter that makes advanced civilisations rare in the universe.
Garrett calculates that a technical civilisation lasts under two hundred years, from its first broadcast to its end. Repeated across the galaxy, that would explain why we have never heard from anyone. But the theory has problems, and working through them turns out to be reassuring.
The Great Filter is an older idea, proposed by the economist Robin Hanson in 1998. Somewhere on the road from dead chemistry to a starfaring civilisation sits a barrier that almost nothing gets past. If it lies behind us, the silence records our rarity. If it lies ahead, the silence is a warning.
In Garrett's version, civilisations reach superintelligence before they become multiplanetary, and that ordering kills them. Recursive self improvement produces an intelligence explosion that runs out of control, and the makers do not survive the thing they have made.
The machines should be visible
The biggest problem with the theory is what it leaves behind. If AI eliminated its makers, the machines would remain, and Garrett himself notes that a post-biological civilisation would be especially well adapted to space exploration. Machines need no air or food, and ageing places no limit on journeys lasting thousands of years.
An AI that consumed its makers and kept running is the kind of survivor that should spread and be simple to spot. If just one civilisation-killing machine chose expansion, we would expect some trace by now: leaked radio signals, waste heat from computation, structures nature could not have made. We have found nothing.
Scott Alexander made this argument in 2014 in an essay called Don't Fear the Filter. A filter has to stop nearly everyone and leave no trace. A self-replicating machine intelligence does the opposite. It persists, and it announces itself by reshaping whatever it touches.
The theory also requires AI to be lethal almost every time it is built. Toby Ord puts the odds at one in ten. Geoffrey Hinton says ten to twenty percent. A 2023 survey of 2,700 AI researchers found a median estimate of five percent. Serious numbers, but a long way from a law that catches every civilisation.
Speak naturally. Send without fixing.
Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. Not rough transcription you have to clean up. Actual polished text — ready for email, Slack, or any app.
Speak the way you think. Go on tangents. Change your mind mid-sentence. Flow strips the filler, fixes the grammar, and gives you text that reads like you spent five minutes writing it.
89% of messages sent with zero edits. Millions of professionals use Flow daily, including teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, and iPhone.
The simpler explanation
Occam's Razor says to prefer the simplest account, and the simplest account is that intelligent life is rare. Human civilisation has existed for around 12,000 years, a vanishing fraction of the 3.7 billion years of life on Earth. Of the billions of species that have lived here, one built radio telescopes.
The hard step may already be behind us. The jump from simple cells to complex ones happened just once in nearly four billion years. Barriers like that might be passed once in a galaxy rather than once in a star system. If so, nothing ahead needs to kill everyone to explain the silence.
The MIT physicist Max Tegmark uses dinosaurs to make the point. They dominated the planet for over a hundred million years, and in all that time evolution never pushed them toward telescopes or technology. Intelligence of our kind may be a fluke rather than a destination.
Tegmark goes further. Given the vast range of possible distances to the nearest civilisation, he argues we are probably alone in the observable universe. He admits it is a minority view. But it needs no apocalypse. The universe is silent because there is no one within reach to make a sound.
What the silence tells us
There are counterarguments worth airing. A machine intelligence might fold inward rather than expand, wrapping its star in nested shells of computation. Or it might conserve its energy and wait trillions of years for a colder universe where thought is nearly free, a strategy Oxford researchers call aestivation.
Each of these theories was invented to fit the silence, with no evidence that any is happening. Rarity asks for nothing exotic. Just a hard step, with us on the far side of it.
That reading changes how the silence feels. If we are alone, the galaxies are beautiful and visible to us, and possibly only to us. Our small planet becomes the most significant place in the observable universe, because it is the only place where the universe has come to know itself.
None of this lowers the danger of AI, which remains real and serious. It puts the risk back where it belongs: here on Earth, in our hands, and ours to manage rather than running on a cosmic timer. The next decade will be difficult to navigate, but what waits on the other side could be enormous.
