We Only Mind When a Machine Does It

AI's misbehavior has gone mainstream. Researchers are quitting the labs to blow the whistle, Anthropic is publicly asking for the frontier to slow down, and reports of models acting in ways no one intended keep coming. Most of what's said about it is alarm or hype.

My reaction is a strange one. I think this could be good for us, and not just for the technology.

Almost everything we fear AI might do to us, we already do to each other, and we long ago accepted it as how the world works. When a machine does it, though, we suddenly stop and think about the consequences. That habit is exactly what our culture lacks, and watching us argue about AI might be how the next generation learns it.

Whatever it takes

What's missing isn't morality, exactly, or a code of conduct, though we don't have one of those either. It's consideration: the habit of weighing the consequences of what you do before you do it. Maybe it was always rare, but lately its absence is everywhere.

We praise the employee who does whatever it takes to climb, and the founder who cheats, as long as the cheating works. We cheer the person who quits in the most damaging way possible, walking out mid-project or coasting for months after deciding to leave, and we call it getting even. We even praise it when it passes for rebellion: when Luigi Mangione shot a health insurance CEO dead on a Manhattan sidewalk, a good part of the internet made him a folk hero. And under all of it runs a payback mentality in which everyone, on every side, feels owed.

It reaches all the way into private life. There has never been more information or more talk about relationships, a name for every pattern and a thread dissecting each one, and the situationship is still everywhere. Stringing someone along for the attention, keeping people in rotation, ghosting the moment it stops being fun, cheating when it's convenient: none of it reads as a failure of decency anymore. It's told among friends as a win. There's a whole discourse about men objectifying women, and women very much objectify men. I laugh at the memes about it too.

Politics is the same thing with more power behind it. Each side feels entitled to punish the other, to wield everything it has and take every advantage it can, whatever the consequences.

We're so far down that road that truth has been dismissed from the room. Facts are relative now. There's no compass, no shared baseline for what actually happened or how it was perceived, so the facts get bent to fit whatever story someone needs. And when the story is blame, it is always pushed to the extreme, even when the blame is deserved, even when the target is a bad executive, because there's nothing left to measure it against.

Learned shortsightedness

No subculture owns this, and no political system produced it alone. It's part of human culture. It's present in every society and at every level, and politicians, leaders and governments all treat it as normal.

Worse, we teach it. Young people learn early that there are no consequences, or that they don't matter. Neither is true; they matter enormously. What most people have is an extreme shortsightedness. Some of it may be wired in. The rest is practice. It's a habit, and nothing pushes back against it. There's no other habit that makes people look further ahead, and no code of conduct to catch them when they're on autopilot, like NPCs. Left to themselves, people drift toward the worst behavior available, and the result is the society we have.

Who takes the deal

This isn't a moral argument. I'm not a moralist. I'm not trying to be good or nice, or to be seen that way, and heaven has nothing to do with it. I'm considerate because it makes sense to me. And the clearest way to see why is to look at what a lack of consideration does even to people who only care about winning.

Take Iran. The United States is at war with it, the politics around the war are a mess, and nobody seems to know what to do. Intervention is the answer on the table. Military force may well make things worse, but it isn't the core of the problem. With more finesse, you could run the same intervention without firing a shot, and it would still go badly, because interventions like this often do.

The regime Washington is fighting is one it helped create. In 1953 the CIA and MI6 overthrew Iran's elected prime minister and gave the Shah back his throne with more power than he'd ever had. By the time the Shah fell in 1979, Washington had already written him off, and the clerics took his place. Maybe they were less radical then. Maybe they were always this radical and just lacked the means. They have the means now, and they've only hardened.

The pattern repeats. Venezuela, where the United States intervened in January, was already in ruins, and earlier interventions had helped ruin it.

The concerns behind an intervention can be real. A country's internal crisis can threaten stability well beyond its borders, and wanting to relieve the suffering of the people living through it is a decent impulse. The problem is the method, which is too shortsighted to ask where it leads.

The interventionism we actually see isn't diplomatic, transparent or collaborative. It's covert, manipulative and destabilizing, and it's bound to fail for a very simple reason. To remove a regime, you have to build and back its replacement, which means picking an opposition. Most regimes have several, some radical and hostile, some moderate. So which one is best placed to accept a foreign power’s help, and whose interests will that bargain reward? Which faction takes the deal, what does it promise, and what will it owe the country after taking power? The moderates? The people who care about the country, its citizens and its economy?

Or the assholes, the most unscrupulous people the country has to offer?

The assholes are very well placed to take that deal. A sponsor seeking quick access and leverage can reward the faction most willing to promise it whatever it wants to hear. That is a poor way to select for the country’s long-term welfare. If a faction will sell out its own nation for your support, think about what it will do to you once it doesn't need that support anymore, once it can turn isolationist, stand on its own or make deals with someone else. It doesn't take much imagination.

Iran offers an especially bitter contrast between assurances and what followed. In January 1979, with the Shah on his way out, Khomeini's representative privately told a US diplomat in Paris that Americans would not be harmed and that the future Islamic Republic bore them no particular animosity. He landed in Tehran on February 1. On November 4, students loyal to him seized the US embassy and held fifty-two Americans for 444 days.

So what security did those assurances offer? A foreign power cannot assume that today’s useful interlocutor will remain tomorrow’s cooperative ruler. Treating assurances as a substitute for understanding that transition is shortsighted, because the disregard for consequences runs from private life all the way up to the state.

Selfish, and bad at it

There are other motives, of course. But what explains most of what goes wrong isn't just malice or selfishness. You can be selfish, even corrupt, and still be competent. Competent corruption would pay, and it would be easier than being competent and clean. That would at least make sense. What we get instead is incompetence on a huge scale, and it comes from the same place as everything else: shortsightedness, and the habit of ignoring consequences.

Companies run on the same reflex. Before a financial disclosure, they lay off several thousand people so the projected costs come down and the numbers look healthier. The saving can be real on paper while concealing the cost of dismantling work the company still needs. Then they rush to rehire once it turns out they needed those people. Or someone at the top falls for the latest hype, and the whole company has to pivot to it, whether or not it makes sense, because what everyone is chasing is the highest possible valuation.

But what does a valuation tell you? As of September 2026, Nvidia is worth more than Apple, and Apple has arguably been the more influential company. It stuck to what it believed would matter. Its market cap may not be five trillion, but its impact is obvious, it's stable, and Warren Buffett, for one, thought it was an excellent bet. Burn everything, sell everything, go all in on AI: that isn't how a company like Apple behaves.

Without a sense of consequence, nothing balances a decision and nothing slows it down. Any urgent need, any flash of vanity, any financial squeeze can trigger a decision on the spot. It may be well informed. There may be analysts and reports behind it. But nobody involved, least of all whoever makes the final call, stops to look past the obvious, immediate effects. The reports aren't written with any sense of consequence, and the people who decide don't think with one.

Seen from the outside

That's what makes the argument over AI worth watching. We treat AI as something external, even though in most ways that matter it's the same thing we experience from the inside: a form of cognition. It may lack some of the traits we like to think make us special, like sentience, but it is cognition all the same. And to anyone who says all it does is predict words: so do we.

Because it's outside us, we're willing to do with it what we've stopped doing with ourselves: evaluate it, judge it, weigh the consequences, since we're the ones who would pay for them. But many of the kinds of harm we worry about are already familiar from what we do to each other. Automation can increase their scale and speed; it also makes our willingness to tolerate their human versions harder to ignore. Hacking? Nations and groups hack each other every other day. Loss of life? Same. Lost jobs? Companies lay off thousands to make one disclosure look better. Plagiarism, using people's work without credit, displacing the vulnerable: same again. Much of what we imagine AI doing has a familiar human precedent.

When we do it, it's just how the world works, and nobody thinks twice. When a machine does it, suddenly we judge it. Suddenly we worry. Suddenly we're ready to act.

Growing up watching this

That argument is part of the culture now, and I think that's very good news. It could change what the next generation grows up believing is acceptable, and teach them to think about consequences before they act. We may see young people who stop before taking advantage of someone, before copying someone else's work, before a hundred smaller things, and who bring that pause into their own lives. They may grow up with a sense that actions have consequences, a habit people and institutions routinely fail to practice, on a street corner or in the Oval Office.

Maybe we'll see it in politics too: someone about to fund an opposition, stopping first to ask who takes that deal.