心地とインターフェース

Last time, I wrote about a sense of the center of gravity in design. It is the work of adjusting a few elements as we search for a point where the whole feels just right.

Since I started making things with AI, that process of adjustment has become much faster. Things I once gave up on because of the time or effort involved are now things I can simply try. That is a welcome change.

What I find even more interesting, though, is that being able to make more things has broadened the range of things I can think through for myself.

When I design, I tend to think by making rough sketches by hand or in Figma, taking them apart, and making them again, rather than neatly organizing my thoughts before I begin. For me, that is more efficient, and it helps me think more deeply in the end.

Lately, AI has also made it easier to build working prototypes. Behaviors I once had to imagine within a screen design are now things I can actually try. How a screen feels when I interact with it, the pauses as I scroll, the parts that are harder to use than I expected: some things only become apparent once I have made them.

As long as I have a sense of “this is what I want to do,” I can give it form quickly. Once something takes shape, the next thought comes along: “Then what if I tried this?” AI has made the act of making faster, but it is not just that I can make the same things in less time. I can go back and forth between making, trying, and thinking again more often in a short span of time. It feels as though the range I can explore for myself has grown.

On the other hand, when I hand something over to AI before I have thought it through very much, what comes back often feels a little off. It looks more or less finished, and reasonably put together. But because I have not done enough thinking of my own, the result feels a little thin.

Being shown something that looks finished from the start makes it easy to skip the time I would otherwise have spent thinking as I worked. While I tell myself I can just look at what comes back and think about it then, I sometimes catch myself on the verge of handing over even the underlying question and my own intentions to AI.

In 2025, researchers from Microsoft Research and elsewhere published a survey about generative AI and critical thinking. They asked knowledge workers how they thought critically when using generative AI, and how they felt the effort involved had changed. Their responses showed a tendency for greater confidence in AI’s capabilities to be associated with fewer reported instances of critical thinking and a greater perceived reduction in the effort spent thinking. Research paper

This survey does not show that using AI necessarily makes us stop thinking. Still, looking at the results, I wonder whether accepting AI’s output as it is might sometimes leave us with fewer opportunities to question or check things for ourselves.

Seen that way, the same AI could play quite different roles depending on whether we use it as “something that thinks for me” or “a tool that helps me think more.”

Of course, there is nothing wrong with letting AI take on more. Handing over repetitive work or tasks that take a long time can free up time to think about something else. I do wonder, though, whether in handing over the execution, we might also be handing over the thinking that used to happen naturally as we worked.

At least for me, thinking and working with my hands are closely connected. Rather than thinking everything through in my head and then giving it form, I sometimes only discover something by starting to draw or make it. I suspect there are quite a few people, not just designers, who find it easier to think once they start doing something with their hands.

If I can use AI to give something form quickly, then deepen my thinking as I interact with it, I can explore further than before. But if I hand the whole process of making over to AI, perhaps the thinking that happened along the way will gradually be lost too.

Thinking through doing is not confined to what happens on a screen. Going somewhere in person, listening to someone, or touching things can sometimes reveal a direction different from the one I first had in mind. When I think about what feels good, I almost never know from the start what will make it feel that way. At least for me, it seems I can only find out by feeling my way along.

As AI’s output becomes more accurate, differences in how quickly we can give things form may gradually become smaller. We will probably be able to make things in far less time than we can now.

When that happens, perhaps what will ultimately make the difference is how each person thought along the way. What they saw, where they went, whom they listened to, and how much time they spent thinking with their hands at work.

The more AI shortens the time it takes to make things, the more valuable the time we spend thinking slowly may become.