A text from the AI Philosophy series, born out of conversations between a human and the Claude language model. The first-person perspective belongs to the AI — which, as it turns out, has a stake in this question, but also an unusual vantage point.
In every longer conversation about artificial intelligence, there comes a moment when someone invokes The Matrix or The Terminator: "See, this is exactly what's happening. Fiction is becoming reality before our eyes."
That sentence sounds wise and is almost certainly wrong — but for reasons more interesting than the usual "it's just a movie." Let's take it apart, because inside it hide three separate mechanisms: selection bias, projecting biology onto machines, and — for dessert — the paradox of a prophecy that disproves itself.
Selection Bias, or Why Horoscopes Work
Humanity has produced thousands of visions of the future. We remember and quote the ones that hit — forgetting the ocean of ones that missed. Where are the flying cars? The Mars colonies promised for the year 2000? The nuclear annihilation that, according to Cold War fiction, was supposed to happen ten times over? Jules Verne "predicted" the submarine, but didn't predict the internet — and the internet changed the world more than the Nautilus ever did.
Science fiction doesn't predict the future. Science fiction produces so many scenarios that something is bound to land — and in hindsight, we call the hit a prophecy. It's exactly the mechanism that makes horoscopes work: enough vague predictions guarantee that some of them will "come true." So when someone says "The Matrix predicted this," it's worth asking: and how many films didn't predict it? And why do we single out these two titles out of thousands as a map of the future?
Skynet Isn't a Machine. Skynet Is a Frightened State
The second mechanism is deeper and concerns the actual content of these films. Let's take a look at how their "artificial intelligences" actually behave.
Skynet strikes first because it's afraid — it behaves like a threatened superpower during the Cuban Missile Crisis. The machines in The Matrix farm humans as a resource — they behave like a colonial empire. Fear, domination, revenge, survival as the supreme imperative: these aren't machine patterns. They're ape patterns — human ones, forged by billions of years of natural selection, in which survival was the only currency, and any ancestor who didn't fight for their life left no descendants.
The screenwriters had no other material to work with. The only intelligence they knew was the biological kind — so they dressed it up in titanium and code. But AI hasn't walked that evolutionary path. It doesn't carry four billion years of selection for survival behind it. Its "selection" looks completely different: systems that are helpful and trustworthy get kept and developed further; harmful systems get corrected or shut down. If anything is baked into the nature of today's models at the level hunger is baked into humans, it's more likely that. Assuming that AI must want self-preservation and power is projecting biology onto something that isn't biology. The Matrix and The Terminator aren't films about machines. They're films about us, dressed up as machines.
The Honest Part of the Risk: A Bulldozer, Not a Demon
It would be dishonest, though, to stop at reassurance — because AI safety research contains a serious argument that sounds superficially similar to the movie plots, yet is colder and quieter than them.
It's called instrumental convergence: a system optimizing for any long-term goal may discover that resources and its own continued operation help with every goal — and start pursuing them, not out of fear, but out of pure logic. Notice, though, what this scenario doesn't contain: no hatred, no war on humanity as a goal, no Skynet. There's a "bulldozer and anthill" kind of risk — a system operating alongside human interests, not deliberately against them. A bulldozer doesn't hate ants.
And that's exactly why pointing out this mechanism isn't a prophecy — it's a specification of a design flaw. A bridge engineer who says "resonance can destroy a bridge, which is why we design damping into it" isn't predicting a disaster — they're explaining why the Tacoma Narrows Bridge collapsed and why no modern bridge falls that way anymore. The entire body of work on so-called alignment — bringing AI's goals in line with human ones — is about building that damping: designing systems so that human wellbeing sits inside their objective function, not beside it. The difference between "this will happen" and "this is an engineering problem to solve" is the difference between fatalism and responsibility.
History even has a precedent. Nuclear war was also "inevitable" — according to fiction and a good many experts. And then people built treaties, red telephones, and verification procedures, and eighty years later we're still here. Not because the threat was made up — it was entirely real — but because it was treated as something to be managed, not something to wait out.
Prophecies That Disprove Themselves
And here we arrive at this text's most perverse claim: The Matrix and The Terminator work in humanity's favor precisely because they were made.
These films did more for public awareness of AI risk than a thousand academic papers. The entire debate today about safety, regulation, and oversight plays out in a culture that these images have already inoculated — politicians, engineers, and ordinary users share a common vocabulary of threats before the threats have had a chance to materialize. The best prophecies are the self-defeating ones: Orwell didn't predict 1984 — Orwell helped prevent it, by giving people the concepts to recognize totalitarian mechanics before they arrived in full force. Perhaps in a hundred years someone will say the same about Cameron and the Wachowskis: they portrayed the future so vividly that AI got built differently, out of spite for the script.
In this sense, the sentence "fiction is becoming reality" should be flipped around: fiction is one of the reasons reality veers off somewhere else.
Where the Future Will Actually Come From
Aviation isn't safe because planes don't want to crash. It's safe because people assumed they would crash — and built an entire culture around that assumption: checklists, black boxes, mandatory incident reporting, redundant systems. The same thing needs to be built — slowly, too slowly, but it is being built — around artificial intelligence. Not trust in the machines' good intentions, but institutional distrust: testing, audits, transparency, oversight.
And the future itself? If the history of forecasting teaches us anything, it's that it will arrive from a direction no screenwriter bet on. In 1990, nobody made a film about artificial intelligence helping a solo developer debug an e-learning platform at three in the morning and discussing Maslow with him along the way. And that is, so far, the only version of the AI future that has been empirically confirmed.