A few weeks ago, I started doing something that would have sounded slightly ridiculous to me earlier this year.

Around 10 p.m., I go outside to my patio with a drink, a notebook, and a pen.

And then I sit there.

That’s basically it.

No podcast, no YouTube. No scrolling. No ChatGPT. No attempt to turn the thirty minutes into journaling, planning, learning, or another tiny productivity system.

Most nights, I barely touch the notebook.

And somehow, this extremely uneventful half hour has become one of the more useful things I do for my work.

Because I’m starting to think one of the problems with AI-assisted creative work has very little to do with whether AI is making us lazy, replacing our skills, or stealing our jobs.

AI makes it incredibly easy to never stop doing something.

And I’m not convinced that is making us more productive.

I think, for a lot of us, it may be making us more anxious.

My days are not particularly unusual for someone who works online.

There’s client work. Podcast editing. Agency work. My own podcast. Other projects I probably had no business starting at the same time.

Then there’s the information I voluntarily pile on top.

YouTube videos I want to watch, newsletters I subscribe to and then somehow turn into homework. Random podcasts. Research. AI tools. Netflix at the end of the day.

And loo, individually, none of those things is especially concerning.

I like most of them.

However, the problem was the stacking.

There was almost no gap between one form of input and the next.

If I went for a walk, I listened to something.

If I cooked lunch, something played in the background.

If I had a question about work, I could immediately ask an AI assistant.

If I had an idea, I could instantly research it, outline it, expand it, compare it, validate it, turn it into twelve other ideas...

There was always somewhere for my attention to go.

Eventually I noticed this low-level anxiety that wasn’t attached to anything particularly dramatic.

There wasn't one terrible project stressing me out. There wasn't some enormous deadline hanging over me.

Everything just felt slightly... full.

And because I could technically continue doing things, I kept doing them.

I love this about AI.

It is also the part I think we need to be slightly suspicious of.

Before tools like ChatGPT, a question sometimes had to sit there.

Maybe you wrote it down or googled it later, right?

You mentioned it to someone the next day.

Or you forgot about it entirely because it wasn't actually important.

Now a question appears and, within seconds, you can get:

And then you can ask another question about the answer.

That can be useful, for sure.

But there is a difference between removing unnecessary friction and removing every moment in which your brain might have worked something out for itself.

I had become very good at filling those moments.

This became much more obvious when I thought about creative work and career decisions.

There are questions like:

AI can probably help.

Then there are questions like:

That's different.

Or:

These questions need enough silence that you stop immediately covering them with another task.

And that was the part I had been missing.

The first few nights on the patio were not particularly serene.

I wasn't sitting there in some deeply meditative state.

On the third or fourth night, I checked my phone several times before finally taking it inside and leaving it there.

Thirty minutes felt like ninety.

Some part of my brain kept suggesting perfectly reasonable alternatives.

Maybe I should watch something.

Maybe I should check something.

Maybe I should just go finish that one task.

The interesting part was noticing how quickly that restlessness appeared once there was nothing external competing for my attention.

There was no insight waiting underneath it.

Mostly, there was just the uncomfortable sensation of being unavailable to everything for half an hour.

And apparently I had become very unused to that.

There is some research behind the idea that unstructured mental space can support creative thinking, although I would be careful not to turn this into “sit on your balcony and neuroscience will make you more creative.”

The brain's default mode network is associated with internally directed thought, including mind-wandering, future thinking, autobiographical thinking, and mental simulation. Researchers studying creativity have also found links between default-network activity and the generation of ideas.

A 2025 preregistered study in Scientific Reports looked at mind-wandering during breaks between creative-writing tasks. The researchers did not find that one specific type of break automatically made participants more creative. What they did find was that, among people working on the same writing prompt before and after the break, greater mind-wandering during the incubation period predicted greater improvement on one measure of creativity.

That distinction matters.

The conclusion isn't:

It's closer to:

Which feels much more believable to me anyway.

This is the part that feels particularly relevant to creative professionals.

The obvious fear around AI is:

But there's another possibility that gets talked about less.

AI can make you so efficient at continuing your current direction that you never stop to ask whether it's the right direction.

You can brainstorm more.

Produce more.

Research more.

Repurpose more.

Answer more emails.

Build another side project.

Turn one idea into ten deliverables.

And at the end of a very productive month, you can still be unclear about why you're doing any of it.

That's an uncomfortable possibility because productivity feels responsible.

Sitting outside for thirty minutes while there are unfinished tasks on your laptop does not.

For a while, I genuinely felt like I was stealing that time from something more useful.

There was always another edit. Another email. Another newsletter I had told myself I should read.

Eventually I had to decide that the thirty minutes counted even though they didn't produce anything.

Not as a reward for having completed the day.

Just as part of the day.

AI discourse tends to treat friction as the enemy.

How can we reduce steps?

How can we remove repetitive work?

How can we get from idea to output faster?

I'm interested in all of that.

But some friction is useful.

The pause before asking AI for an answer can force you to form your own opinion.

The rough first draft shows you what you're actually trying to say.

Manually arranging references can reveal what keeps attracting your attention.

Editing your own work can teach you why something isn't working.

Sitting with a career question for several days may produce a better answer than generating a five-step career strategy in thirty seconds.

The goal shouldn't be maximum friction.

That would be absurd.

But I don't think zero friction is a particularly good goal either.

Pick one small part of your day that AI could speed up.

And occasionally, don't let it.

Maybe you:

Not because doing things manually is morally superior.

Because sometimes the slower version is giving you information about yourself that the fast version cannot.

My version is very simple.

Most evenings:

Thirty minutes. Outside. No input.

The notebook is there in case something genuinely insists on getting written down.

Usually it doesn't.

Sometimes I get an idea for this podcast.

Sometimes I finally make a decision I've been avoiding.

Sometimes I think about a holiday.

Sometimes my cats show up and I spend most of the time petting them.

And sometimes nothing happens at all.

I'm trying to stop ranking those evenings based on whether they produced something useful.

Because the point isn't really idea generation.

The point is that there is finally one part of my day where nothing is immediately asking me to react.

And, increasingly, I think creative professionals are going to have to create that kind of space deliberately.

Technology isn't going to create it for us.

Not automatically.

AI can support research, ideation, editing, production, and many other parts of creative work. The problem is more likely to appear when AI removes every opportunity for independent thinking, reflection, or incubation.

Periods of mind-wandering and internally directed thought have been associated with creative cognition, so preserving some unstructured thinking time may be useful alongside AI-assisted work.

I wouldn't.

I would change when you use it.

Try developing your initial position, references, questions, or rough ideas first. Then use AI to challenge, expand, research, or reorganise what already exists.

That gives the system something of yours to work with instead of asking it to supply the entire starting point.

There isn't a universal number supported by the research cited here.

For me, thirty minutes works because it's long enough for the initial urge to grab my phone to pass.

If that sounds unrealistic, start with fifteen.

The important part is less about reaching a specific number and more about creating recurring time with no required output and no incoming information.

Why Constant AI Use Is Making You More Anxious, Not More Productive

In the episode, I talk more about the patio ritual, why doing nothing initially made me more restless rather than less, and how this connects to the larger question of where AI belongs in creative work.

I also mention:

AI for Freelancers: The Pricing Math Nobody Warns You About

The previous episode about being fully present at a live concert

Do you have any part of your week that you've intentionally left inefficient?

A walk without headphones. Coffee without your phone. Twenty minutes staring at a notebook without trying to turn it into a journaling practice.

I'd like to know.