Creativity in the Age of Generative AI

Photo by Daniil Zameshaev

There’s a story Yuval Noah Harari tells in his (rather dystopian) 21 Lessons for the 21st Century about the emergence of AI- 

In 2017 Google’s AlphaZero defeated the longstanding championship winning Stockfish 8 program at the game of Chess. Stockfish had been trained on centuries of accumulated human experience at the game, and was able to calculate 70 million chess positions per second. AlphaZero by contrast could only perform a meagre 80,000 calculations per second, and was never ‘taught’ strategy by its creators- not even standard openings and their counters.

AlphaZero only learned the game through machine learning, by playing itself. To conquer the world’s most powerful chess engine it only took 4 hours of training. 4 hours vs. the entire accumulation of human knowledge about the game. In Harari’s words- “Because AlphaZero learned nothing from a human, many of its winning moves and strategies seemed unconventional to human eyes. They may well be considered creative, if not downright genius”.

In typical human tournaments the aid of computers is, of course, banned. To monitor this, human moves are assessed for creativity- an exceptionally unorthodox move is seen as not likely to have come from a human mind. Harari finishes his tale with this point-

“At least in chess, creativity is already the trademark of computers rather than humans”

Written in 2018 (i.e. far ahead of the current, chatGPT borne AI wave), this passage has always stuck with me- particularly within the framing of music creativity and the onslaught of generative AI. A constantly discussed dichotomy at the moment is this- will AI just produce a tidal wave of low-quality slop that drowns human artistic expression? Or will it give birth to a fresh creative renaissance- super powering human creators as a partner, muse or even competitor?

AI as a Tool, Not a Threat

Every technological leap in music has carried with it an initial wave of fear. Synthesizers were once derided as “soulless” machines that threatened orchestral musicians, drum machines were dismissed as the death of drummers, DAWs and plugins were viewed as though they would replace producers and mixer engineers entirely (the same is true for consumption as well as creation- Radio, CDs, Streaming…).

Each of these tools has in fact expanded the possibilities of creation. They didn’t erase the human touch, but rather enhanced it. From Kraftwerk to Kanye West, from Brian Eno’s soundscapes to Billie Eilish’s bedroom pop, new technologies have given rise to new genres, new textures, and new ways of expressing the same universal drive: to communicate emotion through music.

Can Generative AI sit firmly in this lineage? It is certainly another tool in the creative arsenal. The quality of its output- whether it becomes wallpaper music or a cultural landmark- currently depends less on the technology itself and more on how humans use it.

If AI is the instrument, then prompts are the performance technique. Just as musicians once had to learn the intricacies of modular synthesis (check the newsletter header image!) or the quirks of a new DAW, creators are now working to master the art of “prompt engineering.”

The ability to shape, guide and refine an AI’s input to get the desired output has quickly become a skill in its own right. It is not enough to say “make me a song like X.” The nuance lies in the detail- the descriptors chosen and the references layered in through the iterative process of testing and refining. This certainly can only come from human experience.

This skill is not limited to music. In software development, for example, entire workflows are shifting towards AI copilots, where the difference between mediocre and outstanding output comes down to the clarity, context and nuance in the prompt. The same will be true in songwriting, production, and even in live performance settings where AI-generated elements could be woven into human-led creativity. Prompt engineering, much like mixing or mastering, will become an essential competency in the modern creative toolkit.

Think of it as a co-writing session. AI can generate melodies, chord progressions, or even entire arrangements at scale- does crafting or curating these require human input? The current discourse lays down the accusation that AI alone lacks the lived experience, the cultural context, and the intuition to know which ideas truly resonate.

The Next Sgt. Peppers?

There is, however, a provocative question we cannot ignore. Much like in the Chess engine paradigm, could AI one day create something so unusual, so unexpected, that it rivals (or out competes) the most revolutionary works of human creativity?

Sgt. Pepper’s Lonely Hearts Club Band, Mike Oldfield’s Tubular Bells, or John Cage’s 4’33”- all were moments that challenged conventions and redefined what music was expected to be. Could AI ever generate a cultural moment of similar magnitude?

Harari’s chess story is illuminating- AlphaZero’s moves were so unconventional that they appeared creative, even alien, to human eyes. Generative AI could hypothetically create something so unlike the history of human music making that it is considered revolutionary. And if it connects with listeners, what does it matter?

Would we celebrate it as genius? Would we dismiss it as “not real art”? Or would we, perhaps reluctantly, accept that creativity is no longer an exclusively human domain?

At its core, the purpose of music is not technical but emotional. The songs that endure are those that stir something within us- joy, sorrow, nostalgia, collective euphoria. If AI-generated music can elicit those emotions, does it matter whether it came from a human?

The underlying concern lies not in the artistic value of AI-generated music, but in the economic structures that surround it. If human creativity is drowned out financially- if licensing, attribution, and revenue flows are not fairly constructed then we risk a world where human artistry cannot thrive (... even less so than it already does today). That is the existential threat. Not that AI will create music, but that human creators will no longer be able to sustain themselves alongside it.

Is it unlikely that AI will replace human creativity entirely? We are consistently seeing that powerful results emerge when humans remain ‘in the loop’. This is true at least in the more administrative tasks that AI is applied to- where there exists a binary ‘right’ and ‘wrong’ scoreboard for results to be marked against. But in a subjective arena? This comes back to the synthesizers- will AI be a superpowered tool-come-muse for creators, or an uber-aggressive competitor with an unnaturally sustained level of output?

In software development, the same pattern is emerging. AI can write vast amounts of code- but developers are still needed to spot inefficiencies, refine structures, and ensure compliance with standards. AI can speed up the work, but expertise remains a non-negotiable requirement.

For music, this suggests a future where AI augments rather than obliterates creativity. The artist who learns how to harness these tools without losing their own voice will have a distinct advantage.

It’s still the Wild West

AlphaZero showed us that computers can already surprise us with genius. Generative AI is beginning to show us the same in creative fields. Whether it produces slop or brilliance will depend less on the technology itself, and more on how we as humans choose to engage with it.

The future of creativity is not a zero-sum contest between humans and machines, but it is going to be interesting, and it is certainly going to be different.