This year's annual 'Festival Of Creativity' in the South of France was marked by the usual tedious adland excess, but this year took place against the backdrop of a creeping existential question: "what does 'creativity' even mean in the age of generative AI?" (Of course, there's another existential question here about whether or not the fine folk, whose jobs in the main exist to persuade people to buy things, are really 'creative' at all, but let's not let those semantic questions worry us too much — after all, they have plenty of other, more tangible worries coming up!)
For those curious about Where The AI And Creativity Debate Is Currently At, the two poles of opinion can be found on X and Bluesky — the former is full of former-crypto-stans sharing AI-generated shorts, rhapsodising about Google's Vue3 model and proudly showing off their large-breasted waifu creations; on Bluesky everything to do with AI is bad, you're a scab and a species-traitor if you use it, and the technology will never be able to create anything of value.
Meanwhile the broadsheets run weekly pleas from world-famous creatives begging the Government to do something to save the arts from the predations of The Machine, lest the siren call of AI-generated books/films/music/art become unignorable and the masses decide that they don't actually care where their content comes from, who made it, and whether a human was involved in the process beyond prompting (the use of the term 'content' feels like its own sign of where this is all going, of course, but that's a separate argument for another essay).
As a 'creative marketing professional' (a designation that will never fail to cause a small shudder, truth be told) you will doubtless have spent countless hours listening to people variously tell you about how AI is going to 'supercharge' the creative industries, 'unlock' potential up and down the value chain, and how agencies and brands alike just need to lean in to this 'exciting new world' and accept The Machine's loving embrace — but then again, lots of the people saying this are being very well-paid indeed to punt that line, and, looking at some of the numbers in agency land over the past couple of years, it's not clear that said embrace isn't just a precursor to a knife in the ribs.
"The Machine can never truly create, that its innately-recursive nature means the truly-new will forever be outside its grasp."
So, what's the answer? AI as creative superpower, or AI as the thing that's going to put everyone who makes films and videos and art and pictures, who writes books and scripts and plays and articles like this, who crafts ads and copy, out of a job forever?
The first thing to say is that no one fucking knows. For every person breathlessly predicting fully automated creative output by 2026 and AI's first Oscar win by 2027 you will find another saying that The Machine can never truly create, that its innately-recursive nature means the truly-new will forever be outside its grasp. There's no guarantee that the current LLM-centric model of artificial intelligence will continue to improve exponentially (growing whispers suggest the possibility of a cul-de-sac and the need to pivot to different underlying tech to achieve the next step along the road to GenAI) — but, equally, the latest GPT, Claude, Midjourney and Flux models, plus the recent launch of Veo3, have taken us quite a long way towards 'absolutely good enough' for the purposes of advermarketingpr at the very least.
In 2024, storied sci-fi author Ted Chiang penned a widely-read piece in The New Yorker with the polemicist headline 'Why AI Can Never Make Art' — the crux of Chiang's argument boils down to "the act of making art is the act of making choices and communicating said choices to an audience; using AI outsources the act of making choices to The Machine; ergo anything made using AI cannot by definition be art". Perhaps unsurprisingly, this line of argument went down very well with the self-defined creative classes, who were reassured to know that their humanity would preclude their replacement by generative AI. But does it stand up?
"What does 'creativity' even mean in the age of generative AI?"
Think about how you use generative AI (and you all do, don't lie, and if you don't then you probably won't be on this nice, all-expenses paid trip next year, so perhaps take a moment to think about that). How do you relate to The Machine? How do you bend it to your will?
Most people's experiences with ChatGPT or Midjourney (other models are available) involve, after initial 'experimentation and vanity prompting' has worn off, engage in a sort of dialogue with AI, giving it an initial question or idea or request for information or imagery or music or video and then refining that question/idea based on the outputs until you arrive at what you imagined, or what you need.
The crucial part of this is that the initial ask came from you. The Machine is not autonomously deciding to pre-empt your questioning and just give you what it 'thinks' you want (remember, The Machine doesn't 'think'); The Machine is responding to you, to what you are asking it - and the questions that you are asking it are based on choices that you have made, decisions that you have taken, directions that you want to explore.
So how does that stack up, Ted?
At its heart, AI — at least in its current incarnation — is a tool, controlled and directed by a user, and any resulting output from the AI is in turn an output which results from the thinking, imagining, reasoning and directing of the human directing The Machine — perhaps in conjunction with it, but, crucially, the idea comes from us, and the human is still in control. As the artist Ed Atkins recently said of his work, recently on show at Tate Britain, "you can make anything, so it becomes about the choice of what to make"; there is no sense in which, on this basis, work created with AI is not 'creative'. If I develop a response to a client brief and then use AI to deliver, whether in part or in total, the work as described in that response, does that obviate the creativity inherent in said work?
You might believe that work that requires a specific skill, a significant time commitment or a lifetime's learning might be intrinsically superior in some way to one which is produced on a production line, that sculpture is inherently 'more' creative than code — and that's your right! I can't stop you! But, equally, this is snobbery. Jack Vettriano painted all his canvases, after all. You might believe that the involvement of The Machine, of 1s and 0s and software, negates human agency, that handing off the heavy lifting reduces the act of making to something like an 'I'm Feeling Lucky!' button for creative labour.
But then you're sort of ignoring a whole swathe of artistic and creative work that's emerged since the 1940s which melds mathematics and code and machine visualisation, and which feels like an antecedent of today's edge-case work using AI. People are making things today using AI which are creative in the purest sense — making work which would have been inconceivable as recently as a year ago, using tools which were the stuff of sci-fi in the pre-Covid era. If you don't think Italian brainrot is creative — the videos, the mythology, the backstories and narrative which have sprung up around it, all initially powered by AI but drawn from the imaginations of kids all over the world and brought to life with new tools to make things which literally couldn't have existed without The Machine — then, well, I think you're probably in the wrong job. If you can't see the brilliance in something like Trisha Code, making sketch shows using AI video with a uniquely human sensibility, or Mario Klingemann's AI art project 'Botto', using machine learning and the social media to create a new (and yet strangely familiar) image of what an 'artist' can be, or the fact that anyone anywhere in the world can go from 'idea in the back of their head' to 'a thing that they can show people' in no time at all, with no budget, and no training, then you're not trying hard enough.
What does this mean for you, though? Well, loads of you are going to lose your jobs. Sorry! If it's any consolation, though, that applies to lots of the rest of us, too (pour one out for the copywriters). If your work involves making banner ads, you probably don't have a business model any more. Equally, though, we're going to see the top end of the market become significantly bigger — after all, there will be no greater flex in the future than to be able to boast of your ability to afford artisanal, all-human content. Taste is going to matter more — the ability to tell what is good and what is not (still, at the time of writing, a human-exclusive quality) will become a differentiator.
While you might think that Hirst, Koons and Warhol before them aren't artists because their works are the result of them instructing a studio full of eager, minimum-wage artisans (aka art students) to, basically, do what they are told (setting out a series of instructions and processes which, if followed, will result in a work being made), it turns out that the art market disagrees with you - and, after all, the only determinant of what is 'art' is the market (sorry, kids!) The practice of storied artists having teams of people working on their behalf to produce work made to their specifications goes back to the Renaissance, and no-one's trying to claim Caravaggio didn't have the creative chops. Does Marcel Duchamp's 'Fountain' not count as art because he didn't craft the porcelain pissoir, or invent the concept of a urinal? Do Peter Blake and Jan Haworth get called hacks because they didn't take the photos for the Sgt. Pepper cover themselves? Meanwhile one in five hits on the Billboard100 contain samples — samples which are chosen by the artists using them, chopped and screwed and fucked and collaged together, but which were created by someone else entirely — do we question their artistry? Do we fuck, because that would be stupid. So why is it different with AI?
Lots of the arguments against the above boil down, at heart, to 'ickiness'. People find the idea of things appearing from the ether… creepy. The fact that every generative AI model has been trained on the entire corpus of information available online, including a significant amount which is subject to copyright, makes lots of people uncomfortable. This is understandable, but also utterly unconnected to the question of whether it's possible to use AI to be 'creative' in a meaningful sense. There are legitimate questions around AI, copyright and fair use, and there's an equally legitimate argument to be made that generative AI, per se, is based on theft. This might well be right! But, equally, it doesn't mean that it's going to go away. Oh, and good luck arguing that in a court of law — it seems reasonably clear that the teams of expensive lawyers employed by the big tech companies are going to be able to argue that there's little material difference between The Machine's ingestion of everything ever made ever and a theoretical polymath let loose in the library of Alexandria when it comes to 'making work based on the corpus of others' work' (and recent courtroom victories for Anthropic and Meta rather back this up although whether Midjourney gets sued into oblivion by Disney remains to be seen).
So, returning to the theme of the opening (AI wishes it could do callbacks like this), why the fear? Is it because, maybe, in your heart of hearts, a lot of you realise that the work you do stopped being 'creative' in any meaningful sense a long, long time ago? Is it because you know that you've been phoning it in, and The Machine really could do it as well as you, because you've not been doing it very well for a while now? AI is an existential threat to the creative industries - if you're not in the top fifteen percent. So, maybe, think of it as a challenge — have better ideas, have better strategy, have more courage, be more human, because, with those qualities, AI can be the best creative partner you've ever had. Failing that, I will see you at the jobcentre in 2026.
WRITTEN by Matt Muir
