Last Sunday I went to see Godsmack.

I was in the front section. Not barrier-close, but close enough that the bass was doing that thing where you don't exactly hear it anymore. You feel it somewhere behind your ribs.

And for two hours, I didn't take my phone out.

Not for one photo.

Which I realise sounds embarrassingly noteworthy now.

Somewhere around the third song, Sully Erna was on stage, everyone around me was screaming the lyrics back at him, and I got this very strange feeling that I was 16 again.

I don't mean I remembered being 16.

It was more physical than that.

Same electricity in my chest. Same stupid grin. For a few seconds, current me and teenage me felt like they were standing in exactly the same place.

And then I looked around.

A couple thousand other people were doing their own version of it.

That is the part I've kept thinking about since.

Because I spend a lot of time talking about what AI can generate. Images. Writing. Music. Video. Ideas. Entire workflows.

We're getting remarkably good at generating the thing.

I don't think we're getting any closer to generating what happened in that room.

I can listen to Godsmack at home.

Better audio, probably.

I could play the exact set list. Turn the speakers up. Nobody tall standing directly in front of me. No waiting for drinks. No queue to get inside.

That would technically give me the music.

And it would give me almost none of what I went there for.

There is an old sociological idea that fits this surprisingly well: collective effervescence.

Émile Durkheim used the term to describe the heightened emotional energy that can emerge when people gather around the same activity, symbol or ritual. More recent research has applied related ideas to concerts and other live collective experiences, where emotional synchrony can intensify individual experience.

That's much closer to what I was feeling.

The experience existed partly because everyone else was having it at the same time.

And there was another ingredient that seems small until you remove it:

this version of the night could fail.

A vocal could crack.

Someone could miss a cue.

The band could change something.

The crowd could be dead.

Something unpredictable is happening in front of you, and the people involved have to deal with whatever happens next.

You don't get a regenerate button.

This is where the AI part gets more interesting.

There's a well-established idea in psychology called the effort heuristic: under some circumstances, people use perceived effort as a signal of quality or value.

In the original experiments, participants rated things like poems, paintings and even a suit of armour more positively when they believed more effort had gone into making them. The effect was stronger when quality itself was difficult to judge.

The evidence around AI-created art adds another wrinkle.

One set of experiments took AI-generated paintings and randomly labelled them either human-created or AI-created. The images themselves were the same.

People rated the human-labelled versions more positively on measures including liking, beauty, profundity and worth. Follow-up work suggested perceived effort and the sense that there was a story behind the work helped explain some of that difference.

That doesn't mean people always prefer human art.

Other studies have found people sometimes prefer AI-generated images when the origin is hidden, so the picture is more complicated than “humans automatically make better-looking art.”

Which is kind of the interesting part.

The output can be beautiful.

And then the story of how it came into existence changes what we think we're looking at.

There is a strange difference between hearing a breakup song and hearing a technically perfect simulation of a breakup song.

Part of what you're responding to is melody, lyrics, production.

Part of it is this invisible assumption that somewhere behind the thing was another person who had a life.

A childhood.

A hometown.

An embarrassing relationship choice.

Something.

You use their work as a mirror because you assume there is somebody on the other side of it.

This is why provenance matters so much to me in the AI conversation.

Where did this thing come from?

Who made the decisions?

What did they notice?

What did they reject?

Why did they make this version instead of the five other ones?

AI can increasingly reproduce the final surface.

The origin story is much harder to reproduce because there isn't an equivalent life sitting behind it.

A lot of creative work is still sold through the polished final asset.

Here's the logo.

Here's the edit.

Here's the campaign.

Here's the finished brand.

And we hide everything that happened before that because, historically, clients usually wanted the finished thing.

I'm starting to think that's backwards for an AI-heavy market.

If the final polished asset is becoming easier to generate, the path you took to get there becomes part of the value.

Say you're designing a visual identity.

Keep the ugly first sketch.

Show the direction you abandoned.

Show the strange reference you found while researching the client's industry.

Explain why one idea looked attractive at first and then stopped making sense.

You already did all of that.

Usually you just throw it away before the client sees it.

And yet that is exactly where your lived judgment is visible.

A generated logo can arrive in five seconds.

Your reasoning can't.

There's a similar argument for work that happens in real time.

A strategy call.

A workshop.

A live critique.

Working through someone's brief together rather than disappearing for a week and returning with a finished deck.

Those formats have something generative tools struggle to reproduce in quite the same way: another human being is there, reacting to this specific situation while you watch them think.

There's risk in that.

You might ask a question and I don't have an immediate answer.

I might change my mind halfway through explaining something.

You might show me something that completely changes where the session goes.

Oddly, I think that messiness may become a selling point.

I paid for the Godsmack ticket even though I could hear the songs for essentially free.

The set list wasn't the product.

Creative portfolios have become incredibly polished.

Mine included.

We show clean mockups, clean case studies, concise bios.

And sometimes, after looking at ten portfolios, you have absolutely no idea who any of the people behind them are.

That becomes a problem when technical execution gets cheaper.

If you've got a strange route into your profession, put it somewhere.

The industry you worked in before this one.

The obscure thing you know far too much about.

Why you chose photography instead of design.

The fact that you started editing because you were making terrible music videos at 17.

Whatever is genuinely yours.

There is research suggesting that perceptions of human involvement, story and effort affect how people evaluate creative work associated with AI.

Your biography isn't filler around the portfolio.

It may increasingly be part of why somebody chooses you.

There's one implication here that has nothing to do with clients.

If you have a creative hobby that isn't your job, I think there's an argument for leaving one of them alone.

No AI.

Even if AI could technically make it easier.

If you're learning guitar because you enjoy learning guitar, generating the difficult part for yourself defeats the point a little.

Same with painting. Knitting. Writing something nobody will publish.

Taking photographs on holiday.

The friction can be the activity.

The badly played chord is involved in becoming someone who can eventually play it.

I don't really want to outsource that.

Parts of creative production, absolutely.

AI systems can already generate work people sometimes rate as highly as — or even prefer to — human-created images when they don't know who made them.

The harder question is whether people value creative work purely because of its output.

Research suggests we don't. Perceived authorship, effort, story and human involvement can change how we evaluate the same or similar work.

That's why the future of creative work may depend partly on things that aren't visible in the exported JPEG.

One reason appears to be perceived effort.

People can infer value from how much work they believe went into something, particularly when its quality is difficult to evaluate directly.

There also seems to be an authorship effect. In experiments where identical AI-generated paintings were described as either human-made or AI-made, the human-labelled versions received more positive judgments.

That does not prove AI-generated work has no artistic value. It does show that the story of creation affects the experience of the result.

I'd start by showing more of the work that's currently invisible.

Your drafts.

Your reasoning.

The choice you nearly made and didn't.

The context behind the final decision.

And if your work allows for a live component — workshops, critiques, consultations, collaborative sessions — that may deserve more attention than it gets right now.

People can buy polished output everywhere.

Watching somebody think through their problem is still a different thing.

I Went to a Godsmack Concert and It Explained Everything AI Still Can't Touch

This episode started as me talking about two hours in a crowd and somehow ended up with Durkheim, AI-generated paintings, handmade mugs and why you might want to keep your ugly first drafts.

Which, honestly, is fairly representative of this podcast.

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

AI for Freelancers: The Pricing Math Nobody Warns You About