What I learned at Gen Jam Part 1: AI's Jazz & Why True Storytelling Can’t Be Generated

Last week I attended a Gen Jam in Amsterdam and helped organise another one in Hilversum, just before the summer holidays. Somehow, we pulled it off in a few days, powered by community, curiosity, and uh… lots of caffeine.

Honestly? I expected a fun, creative experiment. I’ve been dabbling in AI for the past three years, so I thought I had the basics down. But Gen Jam showed me that to really work well with AI, you need curiosity + critical thinking + creative friction. You need to get hands-on. Try things. Break them. Get it wrong. That’s the only way to really figure out how this new kind of tech actually behaves.

Because let’s be clear: AI isn’t just a better search engine or a faster production tool—though we often default to using it that way. Generative AI is fundamentally different. And working with it feels… different too.

What Gen Jam Triggered in Me

In just a few hours, I got confused (so many tools!), excited (look what I can make!), nearly cried (thanks to stunning storytelling from the teams), and had my patience tested (generation takes time and moderation can be brutal).

It also made me pause and ask some other questions:

What does it mean to tell good stories when you can generate anything?
How does AI open up creation to everyone, but not just anyone?
Why does “thinking before you prompt” matter more than we realize?

And most surprisingly… how AI unexpectedly made space for more human connection. As we waited for AI to generate, we discussed and talked.

Why AI Feels So Different?

Because it IS different from much of the software we are used to working with. Generative AI isn’t deterministic like traditional software and apps, it’s probabilistic. The model is always trying to predict what’s most likely to come next. That’s why the same input can lead to different outputs. It’s guessing, based on what it’s learned.

We had just a few hours to build a short film from scratch using AI tools, from concept to script, to visuals and sound. What and how you prompt matters. But unlike the traditional programs we know and love, generative AI doesn’t do exactly what you want.

You don’t click and create. You prompt, wait, see what happens, adjust, and prompt again. That creates a completely different creative rhythm.

Traditional software is like a calculator or a command-following butler, it does the thing you tell it to do, the same way, every time. You click “crop image,” it crops the image.

Generative AI isn’t like that. Under the hood, it’s calculating and making educated guesses. It’s trained on massive datasets and works probabilistically. That means the same prompt can generate completely different outputs because the system is predicting what might come next, not executing a set instruction. This doesn’t happen with traditional tools. Word processors don’t rewrite your sentence in a fresh style. Video editors don’t toss in surprise moons for fun. Unless now powered by AI of course.

Ask a generative AI for a paragraph or an image twice, and you’ll get two different results. That’s by design. It’s not broken. It’s exploring possibilities, not running a fixed script. Because tools like ChatGPT don’t have long-term memory by default (unless enabled), it might repeat itself or shift tone mid-conversation, ot out of rudeness, but because it literally doesn’t remember unless you help it do so.

Early on, I found this unpredictability annoying. Shouldn’t software be consistent and reliable? But I’ve come to see this as a feature, not a bug. The AI’s “wrong” answer might be the start of a more interesting idea. Its strange image might unlock a better version of your story. It’s more improv theatre than assembly line.

The Jazz Metaphor

Jamming with AI feels a bit like playing jazz. Sure, you can improvise freely based purely on instinct, throwing out prompts and seeing what comes back, riffing off the AI’s last response. You might stumble onto something cool. But true fluency—the ability to consistently craft compelling results and steer the creative session where you want it to go—comes from understanding the underlying ‘music theory’ of the tool. Knowing your ‘scales’, the fundamental prompting patterns and syntax that work reliably. Feeling the ‘rhythm’, the iterative loop of prompt, generate, evaluate, refine. Understanding the ‘chords’, how different techniques can be combined for richer outputs.

It’s not about rigid rules; it’s about internalising the structure so deeply that your improvisation becomes intentional, surprising, and harmonically complex. The better you know this ’theory’, the model’s capabilities, its biases, its response patterns, the more interesting, controlled, and truly collaborative your creative jam session becomes.

But Fluency Alone Doesn’t Guarantee a Great Story

You can be a prompting wizard. You can coax gorgeous visuals from Midjourney and witty dialogue from ChatGPT. But if you don’t understand storytelling, emotion, structure, character, motivation, tension, theme..then your output will still fall flat.

During Gen Jam, the most powerful projects weren’t the most technically polished. They were the ones where teams instinctively asked: Why does this moment matter? What does the character truly want? What feeling are we trying to evoke here?

The AI provided the brushstrokes, faster than ever and giving more opportunities than ever. But the meaning and the feeling? That came from the humans. Both creating and interpreting.

Yes, AI can mimic the arc of a love story or the tension of a thriller. It can recreate the beats, even spark emotions. But it doesn’t feel those emotions. It doesn’t know what heartbreak means. It doesn’t sit with grief, rewrite your playlist after a breakup, or get goosebumps at the exact moment a film score swells.

That’s still us.

Without that human core, even the most dazzling AI output is just noise.

The AI Creative Stack

So what do you need to really work with AI in a meaningful way? For me, it comes down to three core skills:

Know the System
No, you don’t have to become a developer. Just like you don’t need to be a mechanic to drive a car—but you do need to know the basics. Where the brake is. What that flashing warning light means. How to steer. Same with AI: you don’t need to code it, but you need to understand it’s not magic. It’s a machine learning system trained on massive datasets, with quirks and biases and limits. And you’re in the driver’s seat.
Learn the Language
Prompting isn’t just typing. It’s shaping. It’s steering. It’s a craft you practice like playing chords or learning to freestyle. The more you experiment, the better you can jam.
Think With It (and About It)
Just because AI gives you something shiny doesn’t mean it’s right or meaningful. You have to reflect, critique, and sometimes challenge what you’re given. Critical thinking is your superpower.

Mini Challenge

If you’re using AI only for shortcuts, try jamming instead. You might just learn something about yourself too 😉
Pick one story prompt (or write your own): 1. A friendship tested during a thunderstorm, 2. The moment someone realizes they’ve been cloned 3. A breakup text… from the year 2072

Your mission: Use any AI tool you like (ChatGPT, DALL·E, Suno, Runway, etc.) to generate a short scene, poem, song, moodboard or image. Then ask yourself: