June 2026 • 7 min read
Tags: AI, Design Process, Product
TL;DR: There is a growing feeling that if something takes time and you have not optimized it with AI, you are falling behind. But not every effort is unnecessary work. Sometimes the mental process we are trying to shorten is exactly the part that builds understanding, intuition and expertise.
There is this feeling that if something takes you time and you have not optimized it with AI, you are behind.
In a designers’ group, someone shared that their team has six researchers, and with an LLM they conducted a new research project in one hour.
In a LinkedIn post that said something like, “Do not send me things Claude wrote, I can ask it myself,” one of the comments was that if you are not using Claude, you are not making your work several times more efficient.
And that creates pressure.
Suddenly, every process that takes time starts to feel like a sign that we have not advanced enough. If we did not shorten it, automate it, or let a tool do it for us - maybe we are working the wrong way.
The problem is that this is not always true.
When I conduct user interviews or usability tests, I can never understand them properly only through generated summaries and by asking questions about transcripts.
I understand best when I watch things myself. When I see users hesitate. When I notice where they stop. When I form insights, and then update what I thought would happen against what actually happened.
After that, I go to AI and ask it to review the transcripts, suggest ideas, analyze what it found, and sharpen things I may have missed.
At that stage, it can genuinely help.
But by then, I already know what is true and what sharpens things I understood, and what it invented, exaggerated or misunderstood. I know that because I went through the process myself first.
This reminds me of a period when I wanted to read nonfiction books, but did not have the patience for the full book. A lot of the time, it felt like the same idea was being stretched and repeated across too many pages.
Then I discovered Blinkist, an app that summarizes the main ideas from books.
On paper, it sounded perfect. Why read a whole book if you can get the essence in a few minutes?
I abandoned it after a few months, because I realized something I had not expected: repetition is part of the learning process. Getting only the essence as a summary was not enough for me to really learn, or for the idea to settle in my mind.
A book does not only “deliver an idea.” It lets me encounter it from several angles, argue with it, get bored by it, notice where it repeats itself, understand what connects with things I already knew and what actually shifts something in my head.
That part may feel less efficient, but it is part of learning.
In cognitive psychology, people talk about schemas - mental structures that help us organize and understand new information. You can think of them as the internal models we have in our minds for how things work.
According to Piaget, learning happens, among other things, through two processes: assimilation and accommodation.
Assimilation - new information fits into a model we already have. For example, I see user behavior that matches what I already thought, and it strengthens or sharpens my existing understanding.
Accommodation - new information does not fit what I thought, so I need to change the model itself. For example, I discover that users are not getting stuck where I expected, but somewhere completely different, and that forces me to understand the problem again.
This is exactly the part that interests me: when I watch interviews, analyze them, identify the gap between what I thought and what actually happened, and formulate insights myself, I update my schemas. I do not only receive an answer, I change the way I understand the problem.
But when I let an external tool do the entire process for me, I may skip part of that update. The insight may arrive as a polished summary, but the process did not really happen inside me.
There is a concept called cognitive offloading - using external tools to reduce the cognitive effort required from us in a task. This can mean writing a reminder on our phone, using a calculator, navigating with a map, or asking AI to summarize transcripts.
And this is not a bad thing in itself.
We do it all the time, and for good reason. Good tools are supposed to reduce effort, shorten processes and help us work better.
The question is which part of the effort we are offloading.
If I offloaded repetitive work, transcription, organization, initial pattern detection or rewriting - great.
But if I offloaded the part where I meet the material, get confused, compare it with what I thought, change my mind and build understanding - I may have saved myself from the part that mattered most.
There is another useful idea from learning research called desirable difficulties. The idea is that some difficulties during learning may look inefficient in the short term, but can actually improve understanding, memory and transfer of knowledge in the long term.
This connects strongly to working with AI.
We are always trying to reduce effort, shorten timelines and save ourselves mental work. And that is completely understandable. But sometimes the mental work is not a bug in the process. It is the feature.
It is what forces us to pay attention.
It is what helps us build distinctions.
It is what updates our professional intuition.
When we skip it, we may give up deeper understanding, skip schema updates, and reduce how much we learn from the process. And these are exactly the things that build expertise over time.
Think about preparing a presentation about a project you worked on.
On the surface, this is exactly the kind of task we may want to optimize. You need to remember what happened, organize a narrative, understand what worked and what did not, choose what matters, and write it concisely.
But all of those things are not just steps on the way to a presentation. They are part of the thinking.
When we prepare a presentation ourselves, we are forced to organize the story in our mind. We understand where there are gaps. We notice what is still unclear to us. We rephrase what we learned.
And when we present, we always need to understand more than what is written on the slides. The presentation is only the base for the talk. It is a thin version, scaffolding that helps us tell the story.
If we do not understand things ourselves beyond what is written on the slides, we will not know how to answer follow-up questions, expand, adjust or change direction in real time.
I do not think the conclusion is that we should not use AI.
On the contrary. AI can be an excellent tool when we use it in the right place in the process.
The question is not whether to optimize.
The question is what we are optimizing.
There is a difference between giving up repetitive work and optimizing execution processes, and outsourcing processes of thinking and understanding.
One can improve us and our work.
The other may do exactly the opposite.
Sometimes, if something takes me time, it is not necessarily a sign that I am behind.
Sometimes it is a sign that I am still inside the part where understanding is being formed.