Can machines really be creative? Or are they just exceptionally good at remixing human ingenuity? It’s a debate as old as, well, AI itself.

Let’s start with definitions. Creativity, the experts say, is the ability to generate something original and valuable. But that’s where the challenge lies: what counts as “original”? The human brain is hardly a blank slate – it absorbs influences, reconfigures them, and occasionally produces a stroke of genius. Machines, on the other hand, digest data, spot patterns, and churn out an output that is sometimes shockingly… inventive. The difference? Humans tell themselves compelling stories about where their ideas come from. Machines just execute.

Consider the AI-written novel, the algorithmic painting, or the deep-learning jazz improvisation. These works have fooled critics, won competitions, and even moved people emotionally. But are they true creativity, or are they just well-disguised mimicry? Both perspectives have strong arguments in their favor. 

For centuries, creativity has been compared to a flash of insight, a lightbulb moment. A sudden spark in the human mind, seemingly out of nowhere. But here’s where AI gets interesting. Modern language models such as GPT-4, Gemini or Claude, with their billions of parameters, also produce unexpected leaps – an uncanny, almost spooky ability to connect unrelated concepts in novel ways.

Is it intuition? Hardly. But is it different from how humans create? Not as much as we’d like to think. The human brain is just another pattern-recognition machine, only a messier, more emotional one.

A University of Milan study (in which I took part) once classified human creativity into three metaphorical categories:

1. The birth process. Some creative people experience their work as a long, painful gestation. (That’s a very human thing: good luck convincing a neural network to suffer for its art.)

2. Lego blocks. Others see creativity as assembling existing ideas in new ways. AI thrives here, recombining data tirelessly at inhuman speeds.

3. The lightbulb. And then there’s the lightning strike of inspiration, which, surprisingly, AI also exhibits in its own cold, calculated way.

Sure, machines are unlikely to compose Hamlet out of sheer existential impulse; but they can certainly spin an original sonnet from a brew of Shakespearean readings. Does it matter whether creativity comes with a tortured soul or a well-tuned algorithm? That depends on how romantic you are about the whole thing.

Perhaps the real question isn’t whether machines can be creative, but whether we’re willing to call what they do creativity. The knee-jerk reaction is to dismiss AI as an imitator, but human innovation has always built upon itself. No artist creates in a vacuum, no scientist formulates theories without predecessors, and no musician invents an entirely new melody out of thin air.

Maybe the ability to create truly original things (rather than copies) will always remain a unique prerogative of humankind. Or maybe creativity isn’t a divine gift exclusive to humans but just a process – one that machines, in their own misterious way, are increasingly learning to master. Time will tell.

Note:

This newsletter is written in collaboration and constant dialogue with several AI tools, which I describe in my books as “today’s best storytelling assistants”. Nonetheless, all views expressed here are my own.