Generative AI: A Force for Good, Not Fear
TL;DR it’s going to be impossible to hold back the tide of generative AI, but it’s alright. It won’t stop human creativity. Actually, it will mean an even more important role for humans. We’re all artists now. (I also introduce the Cutting-edge Creativity Curve and added references to Barry Manilow, De La Soul and Michelangelo).
AI: Fear & Outrage
Back in the 1970s and 80s, the British Musicians’ Union ran a campaign to ‘Keep Music Live’. The campaign — almost unthinkable now — was a war against recorded music itself. They feared that radio and ‘discotheques’ would (gasp!) play records instead of featuring live bands, thus reducing the amount of work for ‘real’ musicians.
The Musicians’ Union even started a campaign to ban the use of synthesisers. In that fight, Barry Manilow of all people, was first in their sights. Of course he was.
The fear and outrage with which AI-powered creative tools have been greeted in certain quarters, including the recent ‘No To AI Generated Images’ campaign on ArtStation, reminds me of these failed, Luddite campaigns from the 1970s.
But it worked out OK. While there may be fewer symphony orchestras, or large recording studios now than in 1970, there is much more actual music-making. In fact, recorded music revenues are now at an all-time high of $15bn annually. And numerous artists, including David Gray, Calvin Harris, The Streets, Moby have all had number one hits which were made at home. Billie Eilish wrote and recorded her entire debut album with her brother in his bedroom studio.
Music didn’t end, it just changed, as did our idea of what constitutes a ‘musician’.
And the same is true of AI generated art.
Headlines such as Your Creativity Won’t Save Your Job From AI, AI is rewriting the rules of creativity. Should it be stopped?, Creatives up in arms over claim that AI is killing human art and the recent protest at ArtStation, are just the latest version of the UK Musicians’ Union concerns. And will end the same way.
The new AI Creative Toolset
So what are we talking about here? It’s been widely discussed elsewhere, so I won’t go into it in a lot of detail, but it’s fair to say Generative AI is impacting every form of human creativity.
Large language models like GPT-3, LaMDA, Bloom and WuDao allow you to create well-written text just from short prompts. Even perfectly structured poems, prose and scripts.
Sophisticated images can be created from text using Midjourney, Stable Diffusion or DALL•E.
And tools like Harmonai, Spleeter and Musico are disrupting music production by automatically generating music or pulling apart existing stems, allowing for producers to remix classic tracks — or generate their own.
As for video, Facebook/Meta and Google are getting into this space, with their Make-A-Video and Imagen tools, and of course there are players bringing faces to life like D-ID (full disclosure, I am VP Marketing & Growth at D-ID).
There are reckoned to be over 350 businesses in this space now and not only is it growing every day, it’s speeding up. AI tools are being integrated into all sorts of business applications.
And the quality is becoming amazing. Two years ago, I wrote about DALL•E 1. Back then, it was still seen as giving quite generic outputs and not to the quality of a finished product. Nothing that could be used in the heat of battle. More of a tool to inspire or mock-up ideas.
These days, though the quality might not be as good as a professional can do from scratch, and isn’t yet fully controllable and repeatable, it’s certainly better than what an average human can accomplish. And it’s a lot cheaper and quicker. All the illustrations in this piece were created for almost nothing, using AI-based tools (although 100% of the text was human generated, promise).
Sampling or stealing?
One of the issues is that the content that the algorithms are trained on is scraped from images on websites like ArtStation and DeviantArt. Indirectly, the artists featured on those websites provide the raw ingredients with which others are now making art.
For sure Generative AI raises issues around ownership and IP. A prime example is Greg Rutkowski. The renowned Polish illustrator and concept artist is one of the names most used in the ‘prompts’ that generate artwork. And he has repeatedly lobbied for living artists to be removed from these training sets.
“Right now, when you type in my name, you see more work from the AI than work that I have done myself, which is terrifying for me. How long till the AI floods my results and is indistinguishable from my works?” Greg Rutkowski
And it’s not just big, famous artists it affects. Small illustrators with very specific techniques can find their entire style ripped off. For example, illustrator Deb JJ Lee who recently found that a model had been built just to imitate her work.
While machines can duplicate and imitate, they lack the essential intentionality of art. To understand what is ‘interesting’ or ‘new’ actually requires knowledge and participation in human culture.
But using other people’s work in your work doesn’t necessarily invalidate it creatively. De La Soul’s 1989 debut album, 3 Feet High and Rising was produced in the early days of sampling. It took elements from over 60 other songs, but no-one could accuse it of lacking creativity. Indeed, it is consistently placed on lists of the greatest albums of all time.
Nevertheless, to this day it is tangled up in IP disputes (try to find it on any streaming service). When they made it, the rules around sampling just weren’t written yet. The early days of these kinds of creative shifts always involve a bit of IP argy-bargy.
Like the arrival of the sampler in music, the kinds of commercial issues that surround Generative AI will necessitate a re-shuffling of commercial arrangements. And there are early signs of it happening already, with Adobe’s announcement that their generative AI policy, “prohibits submissions… including text prompts referring to people, places, property, or an artist’s style.” In Stable Diffusion 2.0, Stability AI themselves have removed the ability to copy artist styles.
And, remember, sampling didn’t stop musicians getting paid — quite the reverse — many musicians are now earn a good living on the royalties from samples of their work. More importantly, the art form moved forward, and, in fact, a new wave of creativity was unleashed. Which leads to my final point…
Nothing dies, it just changes
It’s what I call the Kershaw cutting-edge creativity curve:
- New technology comes along which requires less time/energy for better results
- Initial pushback from vested interests, who want to maintain the status quo
- It’s too useful and good, and everyone knows it
- New technology wins
- Result is more and better versions of that thing
- OLD THING REMAINS, albeit in a new way
Radio didn’t kill novel-writing. TV didn’t kill radio. Film didn’t kill playwriting. VOD and streaming didn’t kill filmmaking. Desktop publishing didn’t kill publishing. Synthesisers, samplers and sequencers didn’t kill musicianship.
In fact, far from killing whatever it is supposed to kill (art, creativity, copywriting, musicianship etc) it typically ends up enhancing.
With the barrier to making images, music, text drastically lowered, the things that become important are the higher level creative questions.
Machine art is not new…
Getting help with the mechanical side of art is nothing new either. Michelangelo used assistants to help him paint the backgrounds of the Sistine Chapel, as did Peter Paul Rubens and Rembrandt. In the modern era, Jeff Koons, Andy Warhol, Takashi Murakami, Damien Hirst, and Marilyn Minter have all employed large-scale studios in which their assistants make their works in a kind-of production line. Koons himself admits to employing a staff of 150, telling the Wall Street Journal, “If I had to be doing this myself, I wouldn’t even be able to finish one painting a year.”
Because, in fact, craft and creativity are not the same thing. There is something critical in the creative act which cannot be replaced by machines.
…and it will liberate creativity
Anyone who has made content with generative tools will tell you there is a huge amount that the human user still has to do, even given the best computer and software in the world. Tweaking, selecting, pushing, prompting and promoting. Because, while machines can duplicate and imitate, they lack the essential intentionality of art. To understand what is ‘interesting’ or ‘new’ actually requires knowledge and participation in human culture. Certainly if you want to get anyone else interested in it.
“With increased automation there will be more time and need for us all to be artists”, Brian Eno, musician and visual artist
As Hungarian-American psychologist Mihaly Csikszentmihalyi points out, “creativity is as much a cultural and social as it is a psychological event… what we call creativity is not the product of single individuals, but of social systems making judgements about individual’s products”. (A Systems Perspective on Creativity, 1999)
As I argued in a previous piece, a machine can’t, in itself, be part of a cultural system. It can’t get past the relevant gatekeepers. It’s just a machine, how can it live in the world or sway anyone?
Another thinker who I admire in this space is electronic musician and pioneer of ambient music, Brian Eno, who has used machines to create his music for decades. Eno talks a lot about the need for humans to spot the interesting things that computers do, because the machines themselves can’t realise they’re doing it. He compares the role of the creator in this new space to the role of a botanist.
“When you’re hybridising plants… you see one that comes up that’s different from all the others. It might have a very distinct pattern on it, or combination of colours, and that’s a ‘sport’. And very often that’s the one you’ll try to grow from.” Brian Eno, talking to Adam Buxton 2017
The point being that what computers throw up, no matter how human-looking, still requires humans to give them meaning.
So, the paradox behind all of this innovation around generative AI, is that it will put the onus on all of us to be creators. With the barrier to making images, music, text drastically lowered, the things that become important are the higher level creative questions. What do I want to say? How do I want it to look or sound? Why will other humans even care?
We’re all creative directors and editors now — which, ironically requires greater creativity. Or as Brian Eno pointed out, “with increased automation there will be more time and need for us all to be artists”.
