Nothing Wrong with Being a Fool

May 25, 2026 Published

In June 2005, Steve Jobs told the graduating class at Stanford, "Stay hungry, stay foolish."¹ The line has been quoted again and again in conversations about startups and creativity.

I think it has often been understood as a call to keep challenging things, to refuse common sense, and to believe in a future that is not yet visible.

So what does it mean to be foolish now, when using AI agents has become part of making things, or maybe even a large part of it?

AI agents help us think, compare, shape, and revise with remarkable strength. There are more and more moments where handing something over looks like the rational choice.

But what do I want to spend time on myself?
What do I want to keep near my own hands instead of handing it to AI?

That much, AI does not decide for me.

Some people call that "taste."² Others talk about the danger that, the more we delegate to AI, the more our understanding becomes hollow.³

Both make sense to me.
But those words still left something tangled. The feeling stayed in the corner of my mind like unfinished homework.

What unexpectedly gave me a way to loosen that knot was Hikaru Utada's "Pappaparadise."

In an interview, she said that because "Pappaparadise" is a song with the message that it should be free, the song itself also had to be free.⁴

True to those words, the song has a lightness that seems to laugh off serious meanings before they can harden around it.

One of the lines in the official English translation is, "There's nothing wrong with being a fool."

It is close to "Stay foolish."
But the temperature I feel from it is a little different.

"Stay foolish" has the heat of a fire that is burning right now.

"There's nothing wrong with being a fool," on the other hand, has the warmth of sunlight.

It feels like permission to keep liking what I cannot explain well.

Someone else may see it as extra.
Still, to me, it keeps mattering.

When I spend time on something that may look foolish to others, I try saying it to myself: there is nothing wrong with being a fool.

I want to think about making things in the age of AI from that feeling.

The smarter our tools become, the more we are asked what we find good

Just because something smart is near us does not mean we want to give up our own intelligence.

If anything, because we can now use smarter tools, we are asked more strongly what we ourselves find good.

In the past, there were many things we simply could not make.
Not enough time. Not enough skill. Not enough people. Those constraints formed a kind of natural boundary.

Now that boundary is moving.

Writing prose.
Writing code.
Designing screens.
Making images.
Researching.
Producing first drafts.

Many things can be done much faster than before.

This is a genuinely big change.

Small ideas that used to stay inside us can now be brought outside much more lightly. In that sense, the entrance to making things has widened. I think that is hopeful.

At the same time, when making becomes faster, the act of making itself can become a little harder to see.

Was this really something I wanted to make?
Or did I mostly tidy up what AI gave me?

I do not think we need to draw that boundary too strictly.

Human-made things have always been made from borrowed materials and layered influences. There is probably almost nothing that is made only from the self.

Still, while making something, there are moments when we feel, "this is good," or "this is not it," and put our hands back into the work.

When using AI, I think what matters is not losing that feeling.

What sits behind the phrase AI slop

There is a phrase: AI slop.⁵

It is often used for rough, similar, textureless content generated in large quantities by AI, and for the disgust or caution people feel toward it.

I understand why the phrase is used so strongly.

The internet already has more than enough information.

If more and more things appear that nobody really looked at, nobody took responsibility for, and that were merely generated, that is exhausting.

As making things becomes easier, the time and attention of the people receiving those things may be treated even more carelessly.

I do not think we should dismiss that anxiety.

At the same time, if we decide that making things with AI is itself careless, I feel we lose something there too.

The concern behind AI slop is probably not simply, "I dislike things made with AI."

There are more practical and more urgent anxieties inside it.

For example, there is the fear that as we keep stacking code produced by AI, the maker can no longer explain why it works.

It runs, but we do not understand it. We can fix it, but we cannot grow it. That kind of state can accumulate little by little.

And then there is the question of taste.

If we ask, AI will produce something reasonably polished.

But it will not decide whether that thing is truly beautiful, whether I can stand behind it, or whether I want to keep touching it for a long time. That judgment remains on the maker's side.

So I think that behind the phrase AI slop, there is not only "do not use AI," but a wish not to treat making things carelessly.

Things are not careless because they were made with AI.
Things are not careful just because AI was not used.⁶

The issue is probably less about the tool itself and more about what we are looking at while we make.

Are we making only to take someone's attention?
Only to increase the count?
Or are we trying to shape something we genuinely find interesting so that it can reach someone else?

Even when the same AI is involved, the difference between those is large.

Making what we love carefully

When I say "carefully" here, I do not only mean spending more time.

I also do not mean polishing every detail perfectly, or finishing everything by hand like a craftsperson. Of course, those forms of care exist too.

But when there are countless smart shortcuts, care may also live somewhere else.

Not treating what I like as disposable.

Stopping properly when something feels off.

Looking at what AI gives me until I can say not "this will do," but "this is good."

That is the kind of attitude I mean.

AI answers very intelligently.
Even when we ask vaguely, it gives back something reasonably well shaped.

That is why, if we are vague, we can still keep moving forward while remaining vague.

This is convenient. It is also a little frightening.

Delegating to AI is not bad.

The question is whether, after delegating, judgment remains inside me.

If I can look at what AI produced and say, "this fits my thinking," "this part is wrong," or "this assumption is suspicious," then I am using a tool.

But sometimes, the moment I read AI's output, it becomes my thought as it is.

I receive a plausible answer before making another view of my own. When that happens, the speed of making goes up, while my understanding slowly grows thinner.⁷

Care in the age of AI may also mean noticing that boundary.

There are many small choices in the middle of making.
Which word to use.
What to remove.
Which inconvenience to keep.
What not to explain.
Where to leave room for play.

If all of those choices are smoothed over, the result may be polished, but it can feel strangely far from me.

To make what I love carefully is, I think, to keep those small choices near my own hands.