Why Better Judgment Matters in the Age of AI

Intelligence does not guarantee judgment.

At the turn of 1942 and 1943, Dietrich Bonhoeffer wrote a reflection for three men close to him, all connected to the German resistance. One section dealt with what he called stupidity. He was trying to understand how intelligent people could lose their inner independence under the pressure of rising power. Slogans took the place of thought, and people became less able to form their own position toward what was happening around them.

Bonhoeffer was writing under a dictatorship, in circumstances radically different from ours, and his explanation belonged to those circumstances. The question he leaves us reaches further. What allows intelligent people to stop exercising their own judgment?

We have always leaned on others

Nobody judges everything alone. We trust doctors because we cannot read the research ourselves. We trust newspapers, teachers, friends and colleagues, and we usually have good reasons. A life lived entirely on first-hand judgment would be impossible.

So the real question is how well we rely on others. Whom do we rely on, for what, and under which conditions? Do we notice when we have stopped asking?

In thirty-three years of working with salespeople, managers and executives in more than a hundred countries, I watched intelligent people get this wrong long before generative AI existed. They had the data and the experience, and sometimes the title. They still accepted a forecast because it was well presented, followed a consensus because it was comfortable, or trusted a confident colleague because confidence is persuasive. Having information was never the same as judging well.

More information, harder choices

AI arrived in a world where deciding what to trust was already getting harder. News is continuous, algorithms shape what we see, and certainty is rewarded more than care.

Information is plentiful. The difficulty is deciding what deserves trust, what deserves attention and what deserves action. Which medical claim should I believe? Which financial advice should I follow? Should I take this job, or choose this school for my child?

Judgment is deciding what deserves trust, what deserves attention and what deserves action when the answer isn’t obvious. Sometimes the facts are incomplete. Sometimes the values conflict. Sometimes we can see the consequences clearly and still don’t know what to do. It has nothing to do with business in particular. Every adult does it every day, usually without naming it.

What AI changes

AI did not create the need for judgment. It changed the conditions under which judgment is exercised.

For the first time, immediate, personalized and articulate cognitive help is available to almost anyone. It can research a question, explain it, compare the options, argue each side, summarize the evidence and recommend a course of action. Increasingly it can carry the action out as well. It will keep improving.

None of this requires pressure. In Bonhoeffer’s account, people lost their independence under the weight of power. AI does not need to force itself on us. It earns its place by being useful. It drafts the proposal in a minute, summarizes the meeting we missed and finds the pattern in the data. When it gets something right, we use it again. When it gets that right too, checking the next answer starts to feel a little inefficient.

There is nothing foolish about this progression. If a tool has been right fifty times, spending an hour verifying the fifty-first answer looks like poor use of time. Each step is reasonable, and the better the tool becomes, the more reasonable each step looks.

When trust becomes surrender

Trusting a reliable source is often wise. I trust my pilot, and I would not want to fly with a passenger who second-guessed every instrument. So what separates sensible reliance from something we should worry about?

The difference is whether we still know where our reliance stops. Delegation keeps a boundary we could name. I look at what the system does well, I decide what it handles, and I understand what I am handing over. Surrender happens when that boundary moves a little at a time until nobody can say where it sits. The assistant drafts, then analyzes, then recommends. Eventually it decides some things. No meeting agreed to this, and nobody announced that judgment was no longer required. The habit of asking simply weakened.

We can lose track of what we handed over. We can also lose track of what the system is aiming for.

Delegation is a choice. Surrender is a drift.

A tool that fails often teaches us to watch it. A tool that succeeds repeatedly teaches us to watch it less. Success changes how we use our own judgment.

What prediction cannot decide

As AI improves, we should give it more of what it does well. That is progress, and I would not want to hold it back. But even perfect analysis cannot settle every decision.

Imagine a future system that can forecast two strategies perfectly. One produces the highest profit and five hundred layoffs. The other produces lower profit and none. The forecast is exact, and the choice is still open. The system can help us think it through and show consequences we would otherwise miss. It cannot decide which consequence should matter most.

The same holds outside business. A person facing a diagnosis, a career change or a question about what to tell their children can get better analysis than ever before. AI can inform the decision, recommend it, and sometimes make it. It cannot live the resulting life on our behalf.

What you own

Responsibility is the right word in an organization, where it means formal accountability. In a life, the plainer word is ownership. You choose the career. You raise the child. You decide whom to trust. You may use experts, friends, data and AI to decide better, and then you live the life those decisions produce.

Suppose a future system understands my finances, history, health and temperament better than I do, and predicts more accurately which choices will go well for me. Why shouldn’t I let it decide? Sometimes I should. That is delegation, and I have argued for more of it.

Now picture it five years later. Every recommendation worked. My income rose, my career advanced, and I never had to agonize over a choice. Then one evening I ask what the system was actually optimizing, and it tells me: income, advancement and the probability of getting hired. All reasonable goals. Along the way, “a better career” had acquired a definition, and I had never stopped to examine it. Maybe I would have chosen exactly that definition. The point is that I never did.

Nothing in that story failed. That is the cost.

“Better” always needs a standard. One job may make me wealthier. Another may give me more time with my children. A third may make my work more meaningful. AI can show me those consequences more clearly than ever before. But what does better mean to me? That is a judgment I want to make deliberately.

Those standards come from somewhere: family, culture, the people around us, and increasingly AI. I want to know which ones are shaping my choices.

A recommendation can be correct for the objective it was given and still be the wrong decision.

Where I apply it

My work concentrates on business because that is where I have spent my professional life, and where consequences are easiest to measure. A salesperson receives a recommendation about which deal to chase. A manager receives a summary of a person’s performance. A recruiter receives a ranked list. A leader receives a strategy. Each recommendation may be well made, and each can be accepted a little more easily than the one before.

The underlying question is the same one every person faces, asked now with increasingly powerful systems helping to shape the answer.

Why Better Judgment

Better Judgment means building the capacity to decide what deserves trust, attention and action, and the habit of choosing deliberately which judgments to keep. Judgment is a capability that grows through practice. Other people can help us judge, and can even decide for us, but they cannot do the practice for us.

AI can help us know more. It can help us see more. It can increasingly help us decide. We should use all of that capability. The risk begins when we can no longer say what we handed over, or whose standard of “better” has been running our decisions.

The decisions AI helps us make still shape our lives, our organizations and the societies we live in.

Bonhoeffer’s circumstances were radically different from ours. His question remains. What allows intelligent people to stop exercising their own judgment? In the age of AI, we have a new reason to keep asking it.

That is why Better Judgment matters in the age of AI.

Source: Dietrich Bonhoeffer, “After Ten Years,” in Letters and Papers from Prison, Dietrich Bonhoeffer Works, Vol. 8 (Fortress Press).