It’s been six weeks since ChatGPT’s new image generator was released. In the time since, the social media posting of AI generated memes soared, although it does appear its died down a bit. First it was Studio Ghibli, then even more strangely, came sharing yourself as an action figure in a box.
I couldn’t help myself.
In the opposite direction, complaints of these trends and what they represent in terms of AI’s challenges to copyright, or downright banality, have also increased.
At times I’ve found this ‘debate’ vitriolic and quite depressing. If all AI content on social media becomes about memes, copyright and the occasional hype about the latest release, then I’d rather switch off entirely.
The polemic
The problem with social media is that it promotes fairly extreme positions. The fed up out of work creative will gain sympathy out of their dismayed anger. It is also easy to appear on the moral high ground against ‘Big Tech’. I found the viral sharing of Studio Ghibli style memes fairly innocuous, there was even a possible PR benefit for the studio, yet no outraged copyright advocate acknowledged this.
Conversely, it is not against the law to copy an artistic style for non commercial work. It is to use that style to form new stories for characters that are protected by IP law and sell them. To state sharing Studio Ghibli style memes is copyright theft flies in the face of the facts. Japan also has one of the most relaxed approaches to AI training in the world.
Even pointing out that it wasn’t OpenAI doing the sharing of the memes was met with anger. I was called ‘a thief’ several times. My own posting about it was pretty on the fence - I’m not really pro AI when it comes to copyright, but at the same time it strikes me that the ability to properly control it in line with current law (to reiterate its not against the law to copy styles) and ethics (which I believe this is actually more about) is unlikely.
But morals and ethics are not the law - they only guide it. I saw an intriguing letter sent to The Guardian along the way:
Authors say they are angry that Meta has used their material to train its Ai (“Authors cal for UK Government to hold Meta accountable for copyright infringement”). But hasn’t that been going on for thousands of years? Isn’t all human thought an iteration of what has gone before? Artists and scientists have been mining the work of others for generations; that’s how human thought evolves.
Ian McEwan was influenced by L.P Hartley’s The Go-Between. George Orwell’s Nineteen Eighty-Four was inspired by Yebgeny Zamyatin’s We. Publishing as a whole is guilty of putting out bandwagon books, which ape the style, themes and trops of a hit. The chief executive of the Society of Authors, Anna Ganley, says writers are “up in arms”. Did she coin that phrase? Creativity has always “trained” on the work of others.
I was surprised by how aligned to my previous thinking this has been. Training is not copying. It is not publishing. There is no law against it.
The training quandary
But there is a serious ethical quandary - an AI system is not human. It is usually owned by a profit making company. It trains on copyrighted work and potentially competes against it. You can see this happening all over the place - with image generators, music generators and video. LLMs will inevitably compete with human written words.
ChatGPT has launched an ‘enhancement’ to its search product that directly competes with commerce focused publishers doing real human research and product testing. It’s difficult for me not to sense the unfairness in this position. Its a position well articulated by Ed Newton Rex in this Ted Talk.
Some creatives want to stop it. Some have embraced it. There is also quite a clear division from what I’ve read about journalism.
Unfortunately for those that don’t want it to progress, this division suggests it won’t be stopped. So long as there are a strong number of advocates then there will be momentum to continue development.
Meanwhile, the overall economic benefits of AI over the long term are going to be too hard to resist for Western governments. They are not going to be able to keep everyone happy along the way.
As the former Managing Director of a production company that also ran its own media, I found AI impossible to ignore. AI refuseniks will simply be much slower at production tasks, which inevitably leads to higher costs vs those who use it effectively. The use of AI is at its most basic a productivity hack, but it will be so ubiquitous that it will lend itself to all sorts of processes.
The increaser?
In my view, AI tools offer enhancement to creative productivity, rather than it being a giant displacer. Many aspects of the creative process are indeed hampered by the technical intricacies of the ‘process’.
For example, in podcast production, fixing low quality recorded audio has been a useful skill for audio editors, but often a tedious one. With that tedious task already somewhat resolved by features of AI tools like Eleven Labs or Descript, audio editors now have more remit to add value in more interesting aspects of production, such as sound design.
Of course, sound design too will be enhanced by AI, but to say an audio editor is redundant because of this folly. The production process will be enhanced, and those who embrace it have license to thrive.
Alongside the enhancement of production processes, new ways of storytelling will be opened up. As an example, visual source material relating to any period before the photograph was invented essentially amounts to paintings and drawings. Some of these are primary sources, but often contemporary illustrators have been commissioned to create representations that fill in the gaps.
Another ChatGPT created image - something which simply wouldn’t have been created without AI.
If you wanted to make a video of a popular folk tale, previously it would be impossible to make without incurring the cost of either doing the artwork yourself or paying an artist.
Well, why don’t you just pay for that? Unfortunately, for the creation of a one minute Instagram Reel or TikTok, which in themselves drives little to no revenue, such expenses are unviable even for publishers with deep pockets. With AI assistance it can be done much more cost effectively.
Artists will understandably have reservations about a process in which they appear replaced – but in this case it is not so much replacement, but enabling viability. Such a piece of content would have likely been unviable. Thus AI allows new forms of storytelling, and artists should look to how they can be involved in such a process rather than rejecting it.
The great video shake up
This has particular ramifications throughout any medium where there are moving images. Of all the televisual processes, animation is one of the most complex and requires a high degree of specialist skill and powerful (and expensive) computers for rendering. It is likely that the creation of visual assets, from illustration to 3D models, will become democratised.
It is already happening at significant scale through Adobe applications, and it is remarkably easy to create assets like a business logo or simple icons using AI. Adding motion and movement to such assets or characters is really the next step and is already available – but doing this to broadcast standard still requires a high level of difficult to learn skill.
Conversely, generative video has the potential to revolutionise the television and film industries. The cost of getting high level production shots of actors in a specific location will likely be further reduced by virtual production – itself a growing part of the film and television production process.
Making video and animation production ubiquitous essentially opens up a creative world where any story can be told much more cost effectively. That will certainly create new rivalries – Disney’s market leading position in animation appears challenged – but it does not spell the end for the industry.
For creators, this can be discomforting. It is quite inevitable. As drones made aerial photography significantly cheaper than hiring a helicopter, AI will reduce the cost of many other forms of creative production.
But none of this really suggests “the death of creativity”, rather a change in the mode and speed of production. The high skills of the future will be about mastering the available AI tools.
Social media and streaming revolutionised the distribution of content. AI will revolutionise its very creation.
Next week I’m taking this debate over to Matt Barr’s ‘Looking Sideways’ Substack, where Alex Roddie, a prominent outdoor writer and journalist is making the argument against AI. It got pretty spicy, so you should sign up.
