We don't lose judgment. We hand it over.

Here is the question that started this for me. If handing decisions to machines is so risky, why do capable people keep doing it, and often eagerly?

The easy answers fall apart on contact. It is not ignorance. The lawyer who filed a court brief full of cases an AI had invented had thirty years of experience. It is not recklessness. The managers who lean hardest on vendor scores are often the most careful people in the building. And it is not some new weakness in the workforce, because the same pattern shows up in every generation that ever picked up a new tool.

The honest answer is less flattering and more useful. We hand judgment over because everything in us is built to. Offloading is one of the oldest human moves. We write the list so we do not have to remember it. We keep the rule of thumb so we do not have to weigh every option from scratch. We trust the pilot, the doctor, the accountant, so we do not have to master everything ourselves. Delegation is not a bug in human behavior. It is part of what made everything else possible.

So the problem is not that we delegate. The problem is what generative AI now lets us delegate, and how easily.

A decision is heavier than it looks

Every decision carries three separate weights, and we usually only notice the first one.

The first is effort. Thinking is work, and the mind treats that work as a cost. We are, as the psychologists have said for decades, cognitive misers. Given two roads to the same place, we take the one that asks less of us. That instinct is not laziness. Attention is a limited budget, and spending it carefully is what lets us function at all.

The second weight is uncertainty. Deciding forces you to sit, for a while, in the discomfort of not knowing. Which reading of this is right? What does this number actually mean? What if I have it backwards? That discomfort is real, and the mind works hard to end it. We reach for the firm answer, almost any firm answer, because closure feels like relief and not-knowing feels like exposure.

The third weight is the one we almost never say out loud. A decision makes you the person who answers for it. When the call goes wrong, someone gets asked why, and being that someone is a cost you feel in advance. It shows up as the knot in your stomach before you commit to the hard thing.

Effort, uncertainty, responsibility. Every real decision asks you to carry all three. And for most of history, nothing could lift all three at once.

The delegate that carries all three

Tools have always lifted the effort. The calculator did your arithmetic. The spreadsheet did your columns. But a calculator never told you whether the number mattered, and it never took the blame if you acted on it wrong.

Human delegates could carry more. Hand a decision to a committee, a consultant, or a deputy, and they absorb some of the uncertainty and some of the blame along with the work. Part of what an organization actually buys from an outside expert is a name to point at when things go sideways. But a human delegate has judgment of their own. They can push back. They can refuse. They can resign. Their name ends up on the outcome next to yours.

Generative AI is the first delegate that carries all three weights at once, and asks for almost nothing in return. It does the work, so the effort is gone. It answers in fluent, confident, well-organized prose, so the uncertainty dissolves the moment you read it. And because there is now a machine in the loop, everyone in the chain has something to point at, so the felt weight of responsibility lifts too. It does all of this instantly, at almost no cost, for any decision you can put into words.

That last part is the one to watch, because it is the one that fools us. The relief is real. The transfer is not. You can feel the responsibility lift, but it has not actually gone anywhere. Courts, regulators, and boards do not accept "the model decided" as an answer, and every year they accept it less. The blame you handed to the system in your gut is still sitting on your desk in the morning. Delegation changes how responsibility feels. It changes almost nothing about where it lands.

Why the wrong things get handed off

If this were simply a story about doing less thinking, it would not worry me much. Offloading effort is usually fine, and often smart. The trouble is what the three weights do to our aim.

We do not delegate the decisions a machine is best at. We delegate the decisions that feel worst to hold. The morally loaded call. The one likely to go wrong. The one that affects other people, where being answerable stings the most. Those are precisely the decisions where a human should stay closest, and precisely the ones the relief tempts us to push away.

Meanwhile we cling to the decisions that flatter us, the ones where our judgment feels like an identity, even when a simple model would plainly do better. The research on this is not subtle. People abandon an algorithm the moment they watch it err, even when it outperforms them, while they hand off the choices that could hurt their conscience. The result is not too much delegation or too little. It is delegation sorted by comfort instead of by evidence, which is close to the opposite of what good judgment would choose.

That is the real crisis hiding inside the AI conversation. Not that we will delegate. We were always going to delegate. It is that we will delegate the wrong things, for the wrong reasons, and feel relieved the whole way down.

---

The next time you are about to hand a decision to a system, try one question. Which of the three weights am I actually putting down here? If it is the effort, and the machine is good at the task, hand it over with a clear conscience. If it is the uncertainty, slow down long enough to notice that a fluent answer and a correct answer feel identical. And if what you are really trying to put down is the responsibility, stop, because that is the one weight the machine cannot take from you, no matter how much lighter it makes you feel.

We will not lose our judgment because machines reached in and took it. We will lose it, if we lose it, because on a thousand ordinary days deciding was heavy and delegating was light, and no one made us choose on purpose.