I genuinely appreciate people who think more than others.
Actually, I take that back.
I appreciate good thoughts.
But thinking doesn’t guarantee good thoughts. Even the smartest person in the world would have stupid thoughts if the inputs or stimuli in their life were wrong.
The thinking crisis
In 2026, we are starting to face a crisis of thinking.
Following the path of least resistance, people will not stop to think twice. The Internet, social media, and brainrot of all sorts were already not helping.
But having AI at your fingertips goes further.
Social media eats up the space for thinking. AI gives you the illusion that you ‘thought about it’.
Delegated thinking
Look, prompting AI doesn’t mean you haven’t thought about it.
It means that for this particular problem, you delegated the thinking to another ‘being’, one that is very smart.
But the scary bit is that your prompt was all the thinking that you did.
We are starting to think in prompts
We must be brutally honest that we are no longer ‘thinking-thinking’.
We are thinking in prompts.
We are thinking about what to write in the chat input, so the AI does the thinking for us. It’s so easy to open the chat interface and delegate the thinking.
Even worse, the best developers in the world are currently designing systems that lack any human thinking whatsoever. Such systems are getting adopted fast because the path of least resistance is… to not think at all.
But for now, this doesn’t mean we don’t think at all. Our minds have started thinking differently. I’ve had prompts that I thought about for 60 minutes. I’ve had prompts that I thought about for days.
So, I do think. I just don’t think about the solution.
My mind is occupied with prompting the AI to solve the problem.
And that might be a problem.
Solving only for the final goal is limiting. The means an AI has of getting there are constrained, because they are purely digital and often lack the context.
Knowing the final goal is not the point. For many real-life problems the journey to figuring it out is the whole point. A human has to walk the walk and involve other humans in the process.
What does it mean to think?
If it’s about the journey (of thinking), then it’s worth focusing on how we walk it.
Thinking is the process of spending time on what you take in. I know three ways to do it.
Thinking before talking
I have friends who think a lot before talking. You ask them a question, and they might think for minutes before answering. I’ve started doing this, too. You are not required to speak within 2-10 seconds of the other person speaking. You can take your time and that’s alright. In fact, that might be much more conducive to the conversation.
Thinking out loud
Many people think out loud. They say the first thing that comes to their mind, and that can work well in friendly conversations, especially if you are not afraid of being wrong. You just spit many thoughts out and see how others react. When I sell a product, I often try multiple variations of my pitch on multiple people and see their reactions. Their reactions show me fast which of my thoughts hold up.
Thinking through writing
Personally, I think best when I write. Having too many thoughts in my head leaves no space for more. Getting them out makes space. And more importantly, I get to review them. I wrote this essay 4 months before publishing it. I changed my mind on some bits and managed to order it better once it had been shelved for a while.
When I write, I mostly spit out my unconscious thinking.
When I review, I do the conscious thinking.
Who are the best prompters? Could they be the best thinkers of our time?
Prompting is a skill that is best learned by… prompting the AI repeatedly. 50 years from now, everyone will be AI-native and will have prompted all their lives. But my worry is that people who grow up prompting AI will default to ‘thinking out loud’ with it, debating with AI at best. They may never learn the other two ways.
An ultra-smart AI will understand even bad prompts, written by people who don’t know how AI works. I think this is already true.
So, how you word the prompt will matter less and less.
What will matter is what you have to say that the AI cannot deduce on its own.
That’s where we say: it’s about judgement, taste, and experience.
But by then, these will no longer be unique. I think there will be ways in which we manage to digitize judgement and experience, while society and culture will probably evolve in some weird way in which machines are good at taste.
What remains for us humans is the actual question. Someone still has to ask it, and a good question provokes thinking instead of ordering an answer.
So, who will be best suited to prompt a superintelligent AI?
I think it might be mentors, advisors, teachers, philosophers, psychologists and similar people, who have worked on developing other people.
People who prompt other people to think.
People who can state a problem without pushing their own answer.
People who know how to ask questions but don’t like giving answers.
People who think about how to prompt other people to think.
If we now think in prompts, these people may also be the best thinkers of our time.
To make this point more obvious: writing essays in school is the definition of being prompted to think through writing. Sometimes it was a completely open question, other times a structure like ‘provide 3 arguments for, 3 arguments against’.
It prompts thinking. That’s why it’s useful for both people and machines.
So?
A prompt can delegate the thinking, or it can provoke it.
Next time someone prompts you to think, don’t pass the prompt on.
Walk the walk. Think the think.

