For most of human history, seeing was believing. A signed confession, a photograph, a recording, or clear video footage of someone in the act—these were the bedrock of proof.
Courts relied on them. So did newsrooms, elections, and the simple daily exchanges where we take one another at our word.
When you pull away that bedrock, the structure above doesn’t just bend. It shatters.
Artificial intelligence threatens to do just that. It isn’t so much an invasion of rogue machines as a steady, quiet erosion of certainty. I think of it as a new religion of doubt—one that, like any faith, asks us to accept its central dogma without needing proof. Its core doctrine? Nothing you see or read can be trusted.
The Collapse of Evidence
Think about how a courtroom actually functions. A prosecutor plays a recorded phone call. A witness points to a face captured on security video. An attorney submits a document bearing a signature.
Every piece of that evidence relies on a quiet assumption: creating convincing fakes is hard, expensive, and usually caught.
That assumption is dead. Today, software can clone a voice from a few seconds of audio. Anyone with a basic laptop can generate a realistic video of someone saying things they never uttered. A fraudster can produce and age a document in the time it takes to brew a cup of coffee.
This danger cuts both ways, but the second path is far more insidious. Yes, innocent people can be framed with fabricated evidence. But guilty people get something just as powerful: the alibi of absolute doubt.
Legal scholars call this the “liar’s dividend.” When anything can be faked, a genuine recording of a real crime can be brushed aside with three words: “That’s AI-generated.” Investigators rely on a chain of custody to prove evidence hasn’t been altered. Now, we suddenly need a chain of custody for reality itself—and we don’t have one.
The Politics of Fog
Democracy depends on a shared set of facts, no matter how fiercely we argue over what to do about them. Partisan media and outrage-driven algorithms have already frayed that shared ground. AI doesn’t start that fire, but it pours high-octane fuel right onto the flames.
Imagine the final weekend before a major election. A video surfaces showing a candidate accepting a bribe, using a slur, or confessing to something terrible. It looks entirely real. By the time fact-checkers can debunk it—if they even can—the votes are already cast.
Now picture the reverse: a real scandal, caught on genuine video, that supporters dismiss as a deepfake while the broader public shrugs in confusion. Both scenarios destroy the same vital thing. Voters stop asking “Is this true?” and start asking “Which side does this help?” Truth becomes a team sport, and belief becomes a matter of tribal loyalty.
Authoritarians have long understood this tactic. You don’t need to convince people to believe a specific lie. You just have to wear them down until they believe the truth is unknowable. A population that throws up its hands and says “who can really say?” is remarkably easy to control.
The Poisoning of Daily Life
This damage won’t stay confined to courtrooms and political campaigns. It will land squarely at the kitchen table.
A grandmother gets an urgent phone call in her grandson’s exact voice, begging for bail money—cloned by software. An employee receives a video call from their boss authorizing a high-dollar wire transfer—executed by a digital puppet. A teenager discovers a fake intimate image of herself circulating among classmates—the image is false, but the trauma is real. A woman receives a deeply thoughtful message online and wonders if a human being actually wrote a single word of it.
Trust isn’t a luxury of modern life; it’s the glue holding it together. Every handshake, contract, friendship, and marriage relies on the assumption that the person on the other end of the line or screen is who they claim to be. When that assumption breaks, we don’t become wiser. We become guarded, paranoid, and isolated.
Doubt at scale is loneliness at scale.
Is This How We End?
That brings us to the hardest question. A giant asteroid ended the age of the dinosaurs. Shifts in climate and volcanic upheavals wiped out most species that ever walked the Earth. Plagues have repeatedly emptied entire cities. Will AI be the engine of our own extinction?
It’s worth looking at both sides honestly.
Serious thinkers—including some of the very people building these tools—warn that superintelligent AI could pursue goals completely misaligned with human survival, or place catastrophic weapons like engineered pathogens into malicious hands. We shouldn’t wave those warnings away. While the most extreme physical scenarios remain speculative, the stakes are far too high to ignore them.
Yet the threat I’m most convinced of is far quieter, and it requires no sci-fi robot uprising. Humanity survives through coordination: sharing reliable warnings about danger, cooperating on a massive scale, and correcting our mistakes together. We beat pandemics and famines because we were able to agree on what was happening and act in unison.
An existential threat—whether a global health crisis, a nuclear standoff, or a warming planet—demands that exact capacity. If AI shatters our ability to agree on basic reality, we lose the primary tool that has always kept us alive. An asteroid doesn’t care about our opinion or our attention span. The dangers of the coming century require both. A civilization that can no longer trust its own eyes might miss the incoming threat entirely because everyone assumed the warning was just a deepfake.
So, will AI cause our demise? It might not be the killer itself. It could simply be the anesthetic that dulls the senses that were supposed to keep us alive.
A Reason to Resist Despair
There is, however, a strong counterargument. Humanity has weathered this kind of shock before.
When photography first emerged, people believed every picture was absolute truth—until they learned photos could be retouched. When the printing press spread, so did counterfeits and political propaganda, prompting societies to build verification institutions: editors, public archives, notaries, and peer review. We adapt, even if the transition is messy and painful.
We can adapt again. Cryptographic signatures can verify the authenticity of a photo or video at the exact moment it’s captured. Legal frameworks can treat digital proof with the same rigor we reserve for physical evidence. Stricter laws can penalize malicious deepfakes, while newsrooms and courts rebuild a modern chain of custody.
Most importantly, we as citizens can cultivate old-fashioned virtues: patience, healthy skepticism, and a strong preference for trusted sources built over years rather than accounts created yesterday.
The difference between healthy skepticism and the religion of doubt is crucial: skepticism asks for evidence, while the religion of doubt gives up looking altogether. One is the beginning of wisdom; the other is total surrender.
The Choice Ahead
Every faith requires believers. This religion of doubt only gains power when we stop checking facts and start shrugging our shoulders—when we decide that since nothing can be proven, nothing matters.
That mindset, far more than any algorithm, is the real threat.
We aren’t helpless against it, but we will have to work to overcome it. The species that survived ice ages, plagues, and catastrophic wars did so because it refused to abandon the distinction between what is true and what is merely convincing. That refusal—not any machine—will determine whether this technology becomes our greatest tool or our final mistake.
Leave a Reply