Death of Reality
If anything on our screen could be fake, what the fuck do we believe?
Written with AI, not by it.
You're scrolling Facebook and there's a chimpanzee and a jaguar, hanging out in the wild like they're best mates. Odd couple, but nature's weird, and for a second or two you marvel at it.
Then something in your brain goes "Wait. That's not just unlikely. That's total bullshit."
Chimps live in Africa. Jaguars live in the Americas. For those two to be mates, one of them would've needed a boat, and the skills to sail it. You don't need a fact-checker or a detection tool (beyond a healthy BS radar) to catch it. You need to know where two animals live, which you learnt way back in the 90s, and it's been sitting in the back of your head ever since, waiting to earn its place. The fake needs you not to know, and in this case, you do. Thank you, David Attenborough.
I've had that exact moment. It was both disappointing and disturbing, and it didn't stay with that one video. The next time an animal does something unlikely on a screen in front of me, suspicion is the default. Maybe it's fake. Maybe it's real and has the bad luck of looking like it could've been faked. From where I'm sitting, the distinction is moot, and either way, the joy I got from seeing the wonders of the world vicariously through my screen is lost.
It'll never be "that's amazing" again. All that is left is "is that real?"
The thing itself
Images have always lied. Wartime posters drew the enemy as monsters, and radio told the home front victory was assured. Magazines were airbrushing bodies, and Stalin was airbrushing whole people out of photos, long before anyone had heard of Photoshop. Fake it till you make it is practically a life philosophy.
But trust was the foundation, and judgement sat on top of it. Life experience and common sense told you the poster was selling you something, and you took the radio with a grain of salt. Even the famous panic over the 1938 War of the Worlds broadcast was mostly newspaper hype. And the lie came dressed as a claim – somebody telling you something. Now it turns up as the thing itself, whether that's a news segment, a jaguar with a new best mate or a feat no ordinary human could pull off, and it looks like it really happened. It comes dressed as the event.
People put unreal things out there and call them real because there's something in it for them: attention, validation, a sale, a vote, a follower. Faking has always needed two things, a tool and a reason, and AI has multiplied both.
Faking used to take a professional, a few days and a real photo to start from. Now it takes a prompt: no skill, images from nothing, open to anyone with a phone, and it can be automated to churn them out by the thousand. What sits on top of reality is a full-spectrum simulation layer. Cheap to produce, profitable to spread, extremely hard to verify at scale. The cost of faking has collapsed. The cost of checking hasn't budged, and someone still has to pay it.
Places that were never there
A Tasmanian tour company's website ran a list of the best hot springs on the island. One, at Weldborough in the north-east, had been made up by the AI writing the company's blog. A group of 24 drivers from the mainland turned up at the Weldborough Hotel asking where the springs were. The publican promised to shout them beers all night if they found them. They never came back. The nearest thing Weldborough has to a hot spring is the Weld River, and it's fucking freezing.
In Malaysia, an elderly couple drove more than 300 kilometres to ride a cable car they'd seen on a news segment. There was no cable car. There was no reporter either, and none of the smiling tourists she interviewed. Afterwards, the wife wanted to sue the reporter.
Travel brochures have been taking the crowds and the trash out of their photos for as long as there have been brochures. A hot spring that was never there is a whole new level. And the doubt doesn't stay with the fake. The list sat Weldborough alongside real springs, so now every hot spring on it has to prove it's there. Once a category's been faked, the real things in it inherit the suspicion. The category flips: fake until proven real.
Photoshop flipped a few categories, narrow ones: nobody presumes a magazine-cover body is real. AI makes every category cheap to imitate, so the flipping has spread into places that until now have been very hard to fake: wildlife footage, news reports, and people themselves, down to their faces and voices. Once enough categories have flipped, you stop presuming reality by default, even in the ones you've never seen faked. Those 24 drivers found out the one way that still works. They went and looked.
Close enough to go and see
Checking it out for yourself is fine for anything that stays put. A hot spring is either there or it isn't. Drive out and you'll know by lunchtime.
Most of what reaches you through a screen doesn't stay put. It happened once, somewhere else, and it's over. Take a clip of a surfer on a wave as tall as an eight-storey building at Nazaré. Until recently I'd believe it on sight. I've seen enough big-wave footage to know people ride waves like that. Even if I hadn't, I could fly to Portugal, stand on the cliff and watch the swell come in, and I'd find out the place is real and that unbelievably enormous waves do indeed break there. But I'd never see that wave.
When Sebastian Steudtner rode the biggest wave ever measured, it took the World Surf League and a team of scientists eighteen months to confirm how big it was. Twenty-six metres, trough to crest. A real wave, with verified footage and the experts on the job, and it still took a year and a half.
So both have work to do now: the real thing has to prove it happened, and whoever's watching has to judge whether it did. Judgement runs on the questions you'd ask that guy telling a suspect story at the pub: did anyone else see it, what's in it for him, what's his track record with the truth, and does it still hold up a week later, once it's been through the wringer that is the internet.
If you'd been on the cliff at Nazaré that day, most of those would have come free. The other angles are the people standing next to you. On a screen, every one of those checks gets harder, and the further away the thing is, the closer they come to impossible, and that's assuming the dude telling the story exists at all.
On no one's authority
An Australian influencer's foundation posted a video announcing a new orphanage in Uganda. The woman announcing it was generated. So were the children, and the banner behind them. Real clips were cut in among the fakes, so even reality came out looking suspect. It took a newsroom's verification team to pull it apart. Aid organisations warned that claims like these wear down trust in legitimate ones. Real children with real needs now have to prove themselves.
In marketing you don't have to look as hard. They want your money. Imagine buying a supplement or a fitness programme on the back of an endorsement from a beautiful, perfectly fit person who doesn't even exist. There is an entire industry of AI influencers out there right now, with followers and brand deals and nothing earned to say.
I saw a clip from a financial adviser speaking from experience, giving advice on wealth generation. The adviser was clearly AI, and the post had even been tagged as AI. The comments underneath were busy praising it. Sure, if the advice was good, it stands on its own. But the authority behind it was resting on a lie. The call to action was buying a wealth-creation course based on hard-won experience that never happened. An imagined roadmap for a real person's dream.
Out of reach
A volcano erupting. A war on the other side of the world. A world leader saying something they may or may not have said. Almost none of us will ever be there for any of it. We have to take someone's word, and it's where the stakes are highest. Getting reliable proof from across the world was always hard. Now there is an abundance of proof for the thing that might not have happened at all.
It's also where the research is. Nobody's going to study fake gardens until they become a bigger problem, but fake politics gets seriously interrogated, and rightly so: people's lives depend on it. When Russia invaded Ukraine, researchers went through nearly 5,000 tweets about deepfakes of the war and found that much of the misinformation came from real footage being labelled fake. Another study showed people in the United States and Singapore a fake video of a local bridge collapsing. Americans came away trusting their government less. Singaporeans, who trust theirs more than people almost anywhere, kept trusting it. The evidence is limited, but what there is points the same way, and it seems to bite hardest where trust was already on thin ice.
So our compass fails in both directions. Believing everything and doubting everything both save us the work of judging, and worn-down people fall into one or the other. Distrust of real news and trusting fakes are two sides of the same coin. The exhaustion manifests as cynicism ("everything's fake anyway"), conspiracy ("we're being lied to about everything") or just not bothering to look closely any more.
And powerful people are taking advantage. In 2024, Donald Trump claimed a photo of the crowd at a Kamala Harris rally had been made with AI. The crowd was real. The claim? Bullshit. Thousands of people had been there, and reporters had photographed them from angle after angle. Thousands more took his word for it.
Democracy assumes we share some reality. Every day we share a little less.
Maybe this is a good thing
There's an animal in Australia with fur and claws that hunts underwater, has a bill like a duck, lays eggs, and the males carry venomous spurs on their back legs. If anyone saw this for the first time today they'd say AI overdid it.
In 1799 a pelt and a sketch of one reached George Shaw at the British Museum. He wrote that it "naturally excites the idea of some deceptive preparation by artificial means", took a pair of scissors to the skin and went looking for the stitches. There weren't any. He put it in his book as a real animal.
So maybe this is a good thing. Trust in images was never really earned. Shaw doubted, checked and got it right, and if everyone has to work like that now, perhaps what's dying is gullibility. Thank fuck.
Except Shaw had one weird animal, a pelt in his hands and as long as he needed. Nobody gets those conditions for everything that crosses their screen in a day. The judging was always there. It used to cost less. Now it's warped by algorithms that feed people whatever confirms what they already believe. Meanwhile the doubt doesn't stay put. It leaks from one category into the next until a real crowd gets called fake, and people don't even ask "is that real?"
If the upshot is that people give up and go out to see for themselves, then that's cool with me. But you can't easily go and see a war, and the things that matter most are often the ones out of reach.
If provenance could be proven, robustly, in a way nobody could fake or buy, this whole argument would thankfully become redundant. But provenance gets faked too. Independent and sustainable labels on food are so loosely regulated they barely mean anything, and on some platforms you can buy the verified tick. Tagging content as AI is more or less an honesty system, and the whole point of a fake is to be dishonest. So it's just another category flipping, one level up. A faked category that stayed contained, without the doubt leaking into the ones next door, would count against me as well. Nobody's shown me one yet.
Lived, not watched
There's one kind of evidence that hasn't flipped. What I see with my own eyes, in the physical world. What I hear from a person's mouth, in the same room, as long as they aren't just repeating something they saw online. Even a voice down a phone line could be a copy. So that's the floor. It's the only place seeing can still be taken at face value.
And the floor itself is eroding. More and more of what anyone knows, and perhaps more importantly believes, about the world now comes from watched material, and the picture in our heads of the universe, the planet and the society we live in gets quietly remixed by systems that have never been there.
And it's generational. If all a kid ever sees is chimps and jaguars hanging out together, where does he learn they've never met, let alone become mates? His teacher can tell him, but can't show him a video of the two of them apart. Given a teacher's word against a video, the kid backs the video. Scepticism only means something when it's got an anchor.
Maybe the younger generations will turn out to be sharper judges than the rest of us. I'd love to see that. I'm not holding my breath.
I'm writing this, and I want you to trust it. You weren't there for any of the things I've described, and many of them I only watched or read about too. Writing from lived experience is itself a category that's been faked. The adviser showed as much. So by my own argument, a real person writing about real experience now carries the burden of proof, or at least has to earn your trust. In comes judgement again. There's no escaping it, and trying to might be what brings us down. In the end, it rests on your shoulders.
Death as we know it
When researchers warned people about deepfakes, the warnings didn't make them any better at spotting fakes. They just made them more likely to call real videos fake. This essay is itself a warning.
So a warning on its own isn't enough. Doubt on its own just flips the coin. You stop believing the fakes and start disbelieving the real thing. The only anchor I can offer is the mechanism. If you can see how the fakes get made, why they pay, and who gains when you believe or when you doubt, you can judge a thing on more than how it looks.
The ability to know what's real goes first at a distance, in wars and speeches and places none of us can get to. Then it goes closer in, as the floor erodes and more of what we believe turns out to have come from a screen. It keeps going until there's nothing left to calibrate the compass against. This is the death of reality.
The world will carry on. Waves will still break at Nazaré. You just won't be able to tell which of the ones on your screen actually did.
Sources
The Ukraine tweets: John Twomey, Didier Ching, Matthew Peter Aylett, Michael Quayle, Conor Linehan and Gillian Murphy (2023), Do deepfake videos undermine our epistemic trust? A thematic analysis of tweets that discuss deepfakes in the Russian invasion of Ukraine, PLOS ONE 18(10): e0291668.
The bridge collapse: Saifuddin Ahmed, Muhammad Masood, Adeline Wei Ting Bee and Kei Ichikawa (2025), False failures, real distrust: the impact of an infrastructure failure deepfake on government trust, Frontiers in Psychology 16: 1574840.
The deepfake warnings: John Ternovski, Joshua Kalla and Peter M. Aronow (2022), The negative consequences of informing voters about deepfakes: evidence from two survey experiments, Journal of Online Trust and Safety 1(2).