That Time I Overheard a Genocide Pitch Between Sets at the Gym
Most people are terrified AI will replace human thinking, but I watched a man casually advocate for nuclear genocide at my gym in Da Nang and realized we handed over our brains to thought leaders long before the chatbots showed up.
I was in Da Nang a while ago, halfway through a mediocre set of pullups, when the guy on the cable machine next to me started explaining why we need to nuke Iran.
Not bomb. Not sanction. Nuke.
“Just turn it into glass,” he said, with the casual confidence of someone recommending a restaurant. “They’re all terrorists anyway. Every last one.”
He was American, maybe late sixties, wearing one of those moisture-wicking golf shirts that cost more than my monthly rent in this city. His conversation partner was British, nodding along with the slightly panicked expression of someone who chose the wrong barstool and can’t figure out how to leave.
What struck me wasn’t the violence of the opinion. I’ve been on the internet long enough to have calluses.
What struck me was how familiar it sounded. The phrasing. The rhythm. The specific cocktail of grievance and certainty.
I’ve heard this exact take before. Word for word. On a podcast. In a tweet thread. Spoken through a microphone at a rally in Ohio.
This man wasn’t thinking. He was playing back.
We’ve Been Outsourcing Our Brains for Decades
There’s this panic right now about AI making us stupid. “People will stop thinking critically,” the argument goes. “They’ll just ask ChatGPT for the answer and accept whatever it spits out.”
And look, I get the concern. I really do.
But here’s what I can’t stop thinking about: we outsourced our thinking to other humans long before the machines showed up.
We just called them “thought leaders.” Or “influencers.” Or “the guy on YouTube who explains geopolitics using video game analogies.”
The man at my gym didn’t arrive at “nuke Iran” through careful study of Middle Eastern history, comparative religion, or even a basic map. He arrived there because someone he trusts told him what to think, and he hit save.
That’s not reasoning. That’s a software update.
And it’s been the default human operating mode forever.
Most people don’t want to do their own thinking. Not because they’re stupid. Because thinking is genuinely hard. It’s uncomfortable. It requires holding contradictory information in your head at the same time without immediately resolving it into a hot take. It means occasionally discovering you were wrong, which feels about as pleasant as finding a dead mouse in your shoe.
So we delegate. To pastors. To pundits. To podcast hosts with deep voices and expensive microphones who sound like they’ve done the reading, whether they have or not.
The AI didn’t invent intellectual laziness. It just made the outsourcing more obvious.
At Least the Machine Has Receipts
Here’s the uncomfortable part of this argument, the bit where I probably lose some people:
If you’re going to outsource your thinking anyway, I’d rather you outsource it to something that can actually show its work.
When I ask a good AI model a question, it pulls from an absurdly large dataset. It’s not a single human with a single ideology and a mortgage to pay. It doesn’t wake up grumpy, doesn’t have a book to promote, doesn’t need you angry because angry people share content and sharing content drives ad revenue.
It has no emotional investment in whether you hate Iranians or love them. It has no political career to protect. It doesn’t care if you join a tribe.
It just… processes.
Now, yes, AI can be biased. The training data has blind spots. The alignment process introduces its own flavor of slant. I’m not naive about this.
But compare the incentives.
A thought leader needs you to keep coming back. Their business model runs on attention, and attention runs on outrage, fear, and the narcotic thrill of feeling like you’re in on a secret the sheeple don’t understand.
The AI’s business model runs on you finding the answer useful enough to keep using it. That’s a very different feedback loop.
Given the choice between outsourcing my thinking to a cable news host who needs me furious by the next commercial break, and outsourcing it to a tool that can at least say “here’s the range of perspectives on this, and here’s the historical context, and here’s what we don’t know,” I know which one feels less like being manipulated.
That Guy Would Have Cheered for Vietnam Too
The thing I keep coming back to, the image I can’t shake, is this:
That same man, sixty years ago, would have been at a different gym, wearing different clothes, explaining why we needed to napalm the Vietnamese. “They’re all communists anyway. Every last one.”
Different enemy. Same playback.
He wasn’t in Da Nang to learn anything about the country whose air he was breathing. He was there despite it. A floating bubble of pre-packaged opinions, drifting through a city that literally proved his entire framework wrong just by existing peacefully around him.
If he’d asked a question or two, if he’d gotten curious for five minutes, he might have noticed that the Vietnamese staff at his gym weren’t terrorists, weren’t enemies, weren’t anything but people trying to get through their shift.
But curiosity is a muscle. And his had atrophied.
That’s what happens when you outsource your thinking to other humans for decades. You don’t just lose your own answers. You lose your own questions.
The Hard Thing
I’m not saying AI is the solution to human laziness. That would be a wild overreach, and I try not to be the guy who makes sweeping claims about technology saving us from ourselves.
What I’m saying is simpler and maybe harder to swallow:
The problem was never about what tool we use. It’s about whether we’re willing to do the uncomfortable work of thinking at all.
Most people aren’t.
And that’s been true since long before the first neural network fired up. It was true when the printing press let people outsource their memory to books. It was true when television let people outsource their worldview to broadcasters. It was true when social media let people outsource their identity to algorithms.
AI is just the latest mirror we’ve held up to ourselves.
If you’re the kind of person who wants to think, really think, you’ll use AI as a sparring partner. You’ll argue with it. You’ll ask it to steelman the position you hate. You’ll treat it like a research assistant who doesn’t sleep, not an authority who can’t be questioned.
If you’re not that kind of person, you were already outsourcing your brain to someone else anyway. The machine is just more honest about what it is.
The man at my gym finished his set, wiped down the machine without being asked (I’ll give him that), and walked away. He never did find out what the British guy actually thought about any of it.
I don’t think he wanted to.