Generative AI: humanity’s creative assistant is here

Image and text versions of generative AI are now available to most people. How will this technology impact the way we create art and write?

In under a year, the explosive growth in generative AI models — machine learning-based systems that replicate human creativity — have boomed. While we had some speakers discussing it at NEXT22 on (appropriately enough) the creative stage, it wasn’t at the heart of our discussion as it would be now.

And in that time, Generative AI has got some people really upset:

It is nonetheless obvious that this industry, by invading the territories of art, has become art’s most mor­tal enemy, and that the confusion of their several func­tions prevents any of them from being properly fulfilled.

Oh, hang on. No. That’s French art critic Charles Baudelaire, writing about the advent of photography in 1859. And now, the best part of two centuries later, his successors are expressing much the same ambivalence about its successor, generative AI.

As The Guardian reported:

An online campaign – #NotoAIArt – has seen artists sharing concerns about the legality of AI image generators, and about how they have the potential to devalue the skill of illustration.

The same piece quotes the illustrator Anoosha Syed:

“AI doesn’t look at art and create its own. It samples everyone’s then mashes it into something else.”

For all their reservations, though, the spread of AI-generated illustrations through newsletters, blogs, and websites is only accelerating. Major AI art platforms like Midjourney and Stable Diffusion are improving at an astonishing rate. And, since the end of last year, we’re also facing an onset of AI-generated text.

The AI as author

ChatGPT’s arrival on the scene has transformed people’s approach to writing already, for good and ill. Many universities are holding meetings or setting up working groups to discuss how to handle assessment in an era of AI-written essays.

One major science fiction publication has had to suspend submissions because they’re facing overwhelming volumes of AI-written submissions. As the editor wrote:

What I can say is that the number of spam submissions resulting in bans has hit 38% this month. While rejecting and banning these submissions has been simple, it’s growing at a rate that will necessitate changes. To make matters worse, the technology is only going to get better, so detection will become more challenging.

AI in academic papers

And it’s not just fiction. Academic journals are rapidly producing policy documents on the use of AI as co-author. For example, Nature has decided on these rules:

First, no LLM tool will be accepted as a credited author on a research paper. That is because any attribution of authorship carries with it accountability for the work, and AI tools cannot take such responsibility.

Second, researchers using LLM tools should document this use in the methods or acknowledgements sections. If a paper does not include these sections, the introduction or another appropriate section can be used to document the use of the LLM.

An assistive creative intelligence

All this, in a matter of months. We’re clearly at an inflexion point in AI’s journey as a technology, when it becomes an “assistive intelligence” in more than one way.

However, this creativity is not uncoupled from human work. The AI doesn’t generate on its own. We write prompts, which the AI uses to generate its work. In that sense, there’s still a human creator making choices, and then leaving technology to do the rest. This is, at least, comparable to modern digital photography, where the human decides how to frame the shot, and when to press the shutter, but the software takes it from there in delivering a final image.

In other words, AI remains a tool for human creativity, not a replacement for it. It is a labour-saving device for art and text, and history shows that humans are always keen to embrace tools that make their lives easier.

Searching for an application

Most prominently, Microsoft has integrated ChatGPT into its Bing search engine, in beta form for the moment. For some queries, it’s surprisingly good. I asked it what NEXT Conference is, and it correctly identified that there are multiple events with the name, and gave me options:

Bing Chat explaining Next Conference

And it didn’t do a bad job of summarising the future of creativity in the age of Generative AI:

That’s a very interesting question. Generative AI is a field of artificial intelligence that aims to design programs capable of human-level creativity. Some experts believe that generative AI can open a new age of creativity and innovation, while others warn of the potential risks and challenges. What are you curious about specifically?

That, my AI friend, is the big question.

Despite my rather benign interactions with it, the Chat version of Bing has proved able to go off the rails in disturbing ways, as Ben Thompson discovered:

I’m not going to lie: having Bing say I am not a good person was an incredible experience (and for the record, I think this is another example of chatbot misinformation!). It also, to say the least, seems incredibly ill-suited to being a search engine. Microsoft (or Google) probably don’t want to be telling their users they are not a good person, and I don’t want to accidentally do a search and miss out on interacting with Sydney!

These sorts of issues have led Microsoft to constrain the beta to six responses per basic query, and 60 responses a day. This is very much a beta, and a long time from being ready. And Google appears to be taking its time to push out its version, Bard, after getting the news out there to prevent Microsoft from taking too much of a PR advantage.

The AI inflexion point

This rush to get a product that clearly isn’t ready for prime-time use in front of the public is a fascinating insight into where we are. Nearly 30 years ago, Bill Gates wrote a memo saying that Microsoft had to get on board the internet wave. They were in danger of missing it, and needed to get their corporate act together. Many tech companies live in fear of that mistake: looking the wrong way, while the next big wave catches them by surprise.

Arguably, that’s been the lesson of the last year as well. While everyone has been talking and thinking about the metaverse and blockchain-based technologies, generative AI has suddenly risen to the top of the heap. Adopt it, or die. And so, we’ll inevitably see lots of the same sorts of clumsy attempts to integrate the tech into existing products we’re already seeing in the coming months. Meta is just the latest to jump on the generative machine learning train.

For example, akeneo has been looking at the potential for well-trained generative AI to improve customer service in online retail:

The use of ChatGPT and other AI technologies is revolutionizing the retail industry by providing customers with omnichannel shopping experiences that are personalized, convenient, and efficient, and that’s exciting. But the big caveat is that your product information needs to be optimized and up-to-date before you can take full advantage of these advancements.

In other words, you need to make sure that the AI is fed with the good stuff — high-quality, accurate product information — to get the best out of it. Fundamentally, the best AI experiences are still built on top of human creativity.

As the Accenture Life Trends report put it:

Human creatives will need to upskill to use the tools, and pay close attention to their craft to ensure quality of output. Companies should think about how they will stand out in the sea of AI-generated content. How can these new AI tools enhance the speed and originality of innovation?

Generative AI is here to stay

What we won’t see, though, is this tech passing away as a fad. Once a tool is invented, it can’t be uninvented. While legal questions about the finished works’ copyright status and the material used to inform them will be fought through the courts, generative AI is here to stay. And anyone in the creative professions will have to adapt, just as Baudelaire eventually adapted to photography:

Although he viewed photography as obviously inferior, and wrote to his mother, “photography can produce only hideous results,” he counted many photographers among his closest friends. He sat for photographic portraits with Etienne Carjat, Charles Neyt and Nadar, and wrote a poem (Le Reniement de Saint Pierre) about clients who visited Nadar’s studio.

Perhaps Generative AI will stay an inferior art. But it will stay. And so, we must adapt.