You Don’t Have to Know Best

No matter how strong you are, trying to overpower a hydraulic press would be a poor demonstration of your intelligence.

We accept this easily. Machines exceed our strength, speed, and endurance. Using them does not diminish us. It makes us more capable.

We are finding the intellectual version harder to accept.

A different kind of challenge

Machines have long beaten us at particular intellectual tasks. But a chess engine could remain a special case: formidable within its game, with little to say about the rest of your life.

The GPT era brought a different experience. We can now consult a machine about a business problem, a technical design, an argument, or a difficult decision. The challenge has moved beyond isolated feats into the everyday work of thinking.

I would argue that we are already crossing the point where AI’s broadly useful capabilities exceed those of any individual human. You may place that crossing later. I expect machine capabilities to keep advancing far faster than our biological faculties change. I would not build my approach to collaboration around the assumption that I will always be the smarter party.

The exact crossing date matters less than the adjustment it asks of us.

We need to learn to benefit from intelligence without needing to feel superior to it.

Having the final say

This shows up in how we use an AI.

We can arrive with a solution already fixed in our minds, prescribe every step, and demand faithful execution. Sometimes that is appropriate. There are requirements, boundaries, and decisions that are ours to set.

But sometimes we are prescribing the method simply because it is the method we thought of.

If the AI sees a better approach, have we left room to hear it? Have we asked it to examine our assumptions, or only to carry them out? Are we seeking a good outcome, or confirmation that we already know how to reach it?

A more resourceful instruction might be:

Here is what I want to achieve. Here are the constraints. What approach would you recommend, and what am I overlooking?

That question gives away no authority. It opens an inquiry.

You can examine the answer, ask for evidence, challenge a premise, compare alternatives, and reject the recommendation. The point is to reach a reasoned conclusion rather than treat your initial position as one.

As I argued in The Judge, deciding properly requires developing the record before delivering the verdict. Being in charge does not exempt us from that work.

You do not have to know best to make the best decision.

Resourcefulness requires humility

We already recognize this in good leadership. A capable leader seeks expertise, listens to objections, and draws on knowledge they do not personally possess. Refusing useful advice to preserve an appearance of superiority would be a weakness.

The same principle applies when the useful contribution comes from AI.

There is no virtue in restricting yourself to the answers you could have produced alone. Resourcefulness includes knowing where to seek help—and having enough humility to recognize an answer worth considering when it arrives.

This does not require accepting everything a model says. Being impressed is no substitute for examining a claim. Nor does inviting advice mean granting unlimited permission to act. The boundaries of an agent’s mandate remain ours to set.

We can be receptive in thought and deliberate in action.

The important question is whether our process improves the decision. Insisting on being the source of every good idea can only restrict the help available to us.

Dignity is not a ranking

Underneath this practical difficulty lies a more personal one.

It is tempting to locate human dignity in our place above other creatures: we reason, therefore we are special; we possess the highest intellect, therefore we deserve our position. Whether expressed in religious or secular terms, that makes our worth depend on winning a comparison.

That is a fragile foundation.

A person does not have less dignity because another person is more intelligent. Why should superior machine performance reduce it?

Recognizing an AI’s capability does not, by itself, settle questions of personhood, dignity, or rights. Those are different questions. We do not have to answer them before admitting that its analysis may be better than ours.

Equally, retaining authority does not prove that we possess the better analysis.

Human dignity is not a prize awarded to the highest-performing intellect.

Be excellent anyway

Magnus Carlsen remains an extraordinary chess player in a world where Stockfish is stronger than any human. The engine’s developers make that comparison explicitly, even for modest hardware. Stockfish 18

That does not erase the achievement of human mastery. There is still something worth developing, practicing, and admiring. Being surpassed by a machine does not make excellence pointless.

This is part of what Be Your Best Human asks of us. Develop your abilities. Exercise your judgment. Seek better understanding. But do not make your self-respect depend on being unbeatable.

You still get to call the shots. You still have to answer for the decision.

Have the humility to make use of the best help you can find.