Why curiosity matters more, not less, as machines get better at answers
It takes a lifetime to understand what to ask. That has always been true. What has changed is that answers have become cheap.
For most of human history, the scarce thing was the answer. We built universities, libraries, consulting firms and whole professions around people who had answers. The expert was the person who knew. Generative AI has turned that around. Answers are now plentiful, instant and getting better. What remains scarce is the question that deserves an answer.
When answers become free, the question becomes the only thing with a price.
So how important will curiosity be in the age of intelligent machines? My answer is that every other skill depends on it.
Let me say why before going further. Prompting is a technique. Curiosity, and the ability to form a question, are capabilities. Techniques expire. I expect the way we talk to a chatbot today to look quaint in a few years, as these systems move into robots and into the ordinary objects of daily life. Capabilities stay with you. The ability to notice, to wonder, to frame, to probe, to challenge and to redirect is not tied to any interface. It is tied to the person.
This is my reading of where things are going, and I may be wrong on the timing. Having said that, I think the pattern behind it is hard to miss.
What matters once answers are easy to get?
The question does. Look at what was scarce in each era, and you can see what people were valued for.
When information was scarce, knowledge mattered. The person who had read the book, met the master or travelled to the city held the advantage. Education leaned on memorisation, because memory was the only store you could carry with you.
When the Internet made information easy to find, knowing was no longer enough. Search mattered. The person who could find the right page, judge whether it was reliable and connect it to the problem at hand moved ahead. A generation learned to search rather than to remember.
Now generative AI has done the same to answers. Not pages to read, but finished answers, argued and formatted, in seconds. What matters now is the quality of the question behind the answer. A vague question gets a generic answer. A precise, honest, well-framed question gets something that can change a decision.
The next step is already visible. AI agents do not only answer. I already see them carrying out tasks across tools, for long stretches, with less and less watching. As action gets cheap, what matters is intention and judgement. What should be done at all? What should never be done?
After that, I expect robots to make physical capability common too: building, repairing, growing, carrying, caring. When capability is everywhere, the scarce thing left is deciding what to do with it.
So the order runs like this: knowledge, search, question, intention, decision. Look closely and the last two are questions as well. What should be done? What should never be done? Each step moves the human contribution away from the mechanics and towards the meaning. Curiosity is what moves a person from one step to the next. Without it, each wave of abundance makes you easier to replace.
Is prompting the skill to learn?
It is a start, and only a start.
People talk about prompt engineering as if it were the frontier. I see it as the kindergarten of a much larger discipline. A prompt is a question you ask a model. But the model is only the first intelligent system we will share our lives with.
Think of an engineer a few years from now, working not with a chatbot but with design agents, simulations and robots on a factory floor. That engineer will not be typing clever sentences. The work will be deciding what the system should optimise. Which assumptions are wrong? Which constraints really matter, and which are only habits? What should the robot watch for? What result would prove our current thinking wrong?
None of that is prompt engineering. It is framing the problem, and it is as old as engineering itself. What is new is that the framing becomes most of the job, because more of the execution is handed over.
In simple terms, whatever the system is, our part comes down to three questions.
What do we want? Why do we want it? What will it take?
The machine can already help a great deal with what it will take, if we ask it. It can help us reason about why we want something. I do not believe it can tell us what we want. Wanting is not a calculation. It comes from where you stand in your own life.
Why does curiosity keep you in the loop?
Because without it, you only approve what the machine proposes.
I used to sit in meetings and wonder: why am I here, and why are we discussing this now? Most of the time there was no good answer. People attend the same meetings and repeat the same work for a simple reason. They never ask why.
That was tolerable when people did the work, because people are slow and someone eventually notices. It is more dangerous when machines do the work. A system that executes without anyone asking why will do the wrong thing faster, at scale and with great confidence.
I can see a loop forming already. The AI proposes. The human accepts. The system executes. In that loop the machine has decided, and the human has only signed. The person who only signs adds little that the machine does not already have. That person is also the easiest to remove.
A person with agency works in a different loop. You observe, you question, you investigate, you challenge, you experiment, you learn, and then you decide and act. Take the curiosity out of it, and what is left is accept and execute.
There is one more reason curiosity matters more now. AI removes friction, and friction used to force us to think. You had to understand a problem to solve it, because solving it took effort. Now you can have a solution without the understanding. That is the trap. The person who stays curious keeps understanding. The person who does not becomes a passenger in their own life.
What changes when a machine can act?
The cost of a poor question.
With a language model, a poor question produces a poor paragraph. You read it, discard it and try again. It costs you a minute.
With a robot, I expect a poor question to produce a poor action in the physical world. On a factory floor, a badly framed task becomes a badly built thing. On a farm, it can become a lost season. The machine does what it was asked, precisely and tirelessly. The outcome will depend on whether the person who framed the task understood the problem well enough.
Think of two ways of working with a robot in a warehouse. The first is a command: clear this aisle. The second is a shared inquiry. Why does this aisle keep getting blocked? Which items are left here again and again? Could we change the layout so that it stops happening? What should you handle on your own, and which decisions should still come to me?
The first makes the robot a tool and the human a supervisor. The second makes the two a team, where the human brings the questions and the machine brings the observation and the labour. The second is only possible if the human is curious.
A robot will do precisely what it is asked. So a great deal depends on who is doing the asking.
Are all questions equal?
No. There is a difference between a question and the right question.
I find it useful to think of questioning as a ladder with five rungs.
The first is clarifying. What does this mean? The second is investigative. Why is this happening? The third is critical. Which assumptions might be wrong? The fourth is generative. What else could be possible? The fifth is transformational. Are we solving the right problem at all?
AI tools can already help with the first three when asked. I find the fourth is where they are strongest. Ask for possibilities and you will get a hundred in a minute.
The fifth rung is different in kind. “Are we solving the right problem?” is not only a question about the problem. It is a question about the person asking: your purpose, and what you are willing to spend your life on.
I think this is where the professions will be rewritten.
A doctor working with AI will not stand out by remembering every disease. The better doctor will ask whether the system is looking at the right problem at all. An engineer will not stand out by calculating every optimisation, but by asking the question that changes the architecture. An entrepreneur will not stand out by writing the business plan, but by noticing the human problem nobody else has noticed. A scientist will have no shortage of hypotheses. The work will be deciding which of them is worth a year.
In every case, the human contribution moves to the fifth rung.
What does it take to ask the right question?
Three things, and they have to start with you.
Knowing what you do not know, which takes honesty with yourself. Knowing what matters, which takes purpose. Knowing to ask what it would take, which is a question about space and not just time.
The third is where most of us fail. When we want something done, we ask when it can be done. That is a question about time, and time runs in one line. We rarely ask what it will take. That is a question about space, not time. It makes you see the dependencies, the people, the unknowns and the things nobody has mentioned yet. And that is where you need your judgement.
I learned this the hard way in years of estimating software delivery. When we missed dates, we blamed the estimates. The estimates were not the real problem. The questions were. We had asked when, and never asked what it takes.
The same applies to working with intelligent systems. Ask a machine when, and it will give you a schedule. Ask it what it will take, and it will show you the shape of the problem. In that shape you will often see the question you should have asked in the first place. I ended my post on why everything that can work, will work on the same note: ask what it would take.
How should we raise the next generation?
By protecting their appetite for the question.
Our schools were built to reward answers. Memorise, reproduce, score. That made sense when information was scarce. It makes much less sense now.
Many teachers know Bloom’s taxonomy, as revised in 2001, which orders learning as remember, understand, apply, analyse, evaluate and create. If I were to redraw it for the coming decades, I would name three more things in it. Questioning, in the middle, because it is what carries a learner upward. Then judgement and agency, at the top. A learner who never questions stays at applying, and applying is something the machine already does well.
The child who will thrive is not the one who knows the most. It is the one who stays curious, and who has learned to turn that curiosity into precise, honest questions. A child growing up with AI should say “give me the answer” less often, and ask more often: Why? How do we know? What would happen if? What are we assuming? What does not fit? What have we not asked yet?
This cannot be taught as a subject on a timetable. It is cultivated. It comes from reading widely, from reflecting, from being around adults who ask why, and from having the time to wonder without being rushed to an answer. I believe it takes a lifetime to understand life, and nobody can shortcut their own “aha” moments. Curiosity is what brings those moments.
So the job of a parent or a teacher is not to supply the answers. The machine does that now. The job is to keep the child asking.
Stay curious
There is a paradox in all this. The more intelligent our machines become, the more valuable curious people become.
Many people expect the opposite. The common fear is that intelligent machines make human intelligence redundant. They would, if human intelligence were mainly about having answers. It is not. AI can widen the search without limit. Someone still has to decide where to look, what matters, what is strange, what deserves investigation and, in the end, what kind of future is worth building. Every one of those is a question.
I do not see the future as a contest between human intelligence and machine intelligence. On answers, the machines may well win, and that matters less than we fear. I think the people who do well will be those who can direct that intelligence. And you direct it with a question.
So if there is one habit to protect and build a life around, it is this one. Stay curious. Ask better questions. Keep climbing to the fifth rung. After all, the machine can know. But we are the ones who have to ask.
Stay positive.
Leave a comment