0
← Journal index

AI / Wanderings 2026

AI Has Made Answers Cheap. Better Questions Now Matter More.

AI can produce an answer in seconds. The real work is choosing a worthwhile problem, challenging the first response and testing assumptions in the real world.

By Martin Uetz5 min read
A person uses a brass compass and a control dial above branching blue paths and one amber route on a wooden table.

AI has made answers cheap.

Give a machine a question and, within seconds, it will produce an answer. The answer may be useful, banal or confidently wrong, but the blank page is no longer blank.

That changes the value of human contribution. Producing an answer still matters. Deciding which problem deserves attention, what a useful answer must achieve and whether the response survives contact with reality matters more.

This is why I keep returning to purpose and curiosity.

Purpose gives work direction when enthusiasm fades. Curiosity stops us accepting the first plausible answer. Together, they help us use AI to extend our thinking without handing over our judgement.

Passion fades. Purpose keeps you moving.

“Follow your passion” sounds wonderful. It also places a lot of responsibility on a feeling.

Feelings change. Some mornings, I am passionate about an idea. By lunchtime, I am passionate about coffee. After three difficult meetings and a broken spreadsheet, the idea has mysteriously lost some of its sparkle.

Passion can supply energy. It is a poor navigation system.

Purpose works differently. It gives direction when the work becomes repetitive, uncomfortable or difficult. You can lose enthusiasm for a task and still understand why it needs doing.

A teacher may feel little passion for marking the twenty-seventh essay on a Sunday evening. The purpose remains: help a young person learn to think and communicate.

A founder may grow tired of chasing invoices, fixing bugs or explaining the same product for the hundredth time. The purpose remains: solve a real problem for people who care about it.

Purpose does not need to be grand. Improving how your team works, helping a customer or making one part of a broken system less stupid can be enough.

It answers a practical question:

Why is this worth continuing when I do not feel like continuing?

Curiosity can be practised

We often describe curiosity as something you either have or you do not. Some people appear naturally curious. Others accept the first explanation and move on.

I think curiosity can be practised.

It begins with a pause before the obvious answer becomes the accepted answer. You notice an assumption. You test it. You follow an unexpected result instead of forcing it back into the original plan.

A curious person is willing to ask:

  • What are we assuming?
  • What evidence would change my mind?
  • Who experiences this problem differently?
  • What happens if we try the opposite?
  • Why are we still doing it this way?
  • What do I not understand yet?

None of these questions requires genius. They require attention and a willingness to look slightly foolish for a few minutes.

That last part matters. Organisations regularly claim to value curiosity while rewarding certainty. The person with an immediate answer sounds competent. The person asking a basic question risks sounding uninformed.

So everyone nods, the meeting ends and the team spends six months solving the wrong problem very efficiently.

AI does not fix that habit. It can make it faster.

Treat the first AI answer as material, not a decision

AI can produce text, images, software, analysis and ideas at astonishing speed. Speed, however, does not decide where to go.

Without purpose, AI helps us produce more things that nobody needs. Without curiosity, it gives us faster versions of familiar answers.

Ask AI to “write a marketing strategy” and you will receive something that looks like a marketing strategy. It will contain audiences, channels, messages and enough headings to frighten a small child. It may still have no meaningful relationship with the customer, the product or the problem.

The useful work begins when you challenge the first response:

  • Why would the customer care?
  • Which assumption in this strategy is weakest?
  • What would make this product a bad choice?
  • How would a competitor attack this position?
  • What information is missing?
  • Where might the model be wrong?

Now the first answer becomes material to examine. You can expose assumptions, compare alternatives and explore paths that would previously have taken days.

The prompt box is a rather unforgiving mirror. A vague question often reveals vague thinking.

Knowledge is what lets you challenge the machine

There is a tempting idea that AI makes knowledge unnecessary. If we can look everything up, why remember anything?

Because curiosity needs something to work with.

Knowledge helps us recognise when an answer feels wrong. Experience tells us which questions matter. Context lets us connect an idea from one field with a problem in another.

AI can offer ten options. A human still has to judge which one fits the situation, the people involved and the consequences of getting it wrong.

The goal is not to outsource thinking. It is to extend the distance our thinking can travel.

That puts more responsibility on us, not less. We need enough understanding to question the output, enough humility to admit our gaps and enough judgement to decide what happens next.

Purpose turns exploration into progress

Curiosity without direction can become endless exploration: interesting, enjoyable and occasionally useful, but endless.

Purpose gives curiosity somewhere to go.

If your purpose is to improve education, curiosity leads towards questions about how children learn, why teachers burn out, what parents need and where technology helps or gets in the way.

If your purpose is to build a healthier company, curiosity leads towards incentives, power, communication, customer behaviour and the awkward gap between the values on the wall and the decisions made in the room.

AI can help investigate those questions. It can summarise material, challenge an argument, simulate perspectives and suggest experiments.

Purpose decides which experiments are worth running.

The next generation will grow up with machines that can answer almost anything. Teaching them merely to produce more answers would be a strange use of their time. They need to learn how to choose meaningful problems, ask better questions, examine what comes back and change their minds when the evidence demands it.

Try this for 30 minutes

Choose one problem you care about. Before you open AI, write down why the problem matters, what you want to achieve and who should benefit.

Then spend 30 minutes asking questions without requesting a finished solution:

  1. Ask the machine to identify and challenge your assumptions.
  2. Request several competing explanations, not one polished response.
  3. Ask what evidence would disprove each explanation.
  4. Identify the weakest assumption in the answer.
  5. Choose one observation you can test outside the prompt box.

Then close the laptop and test that assumption with a real person who experiences the problem.