The Book I Never Published

I’ve used generative AI to create artwork long before it became mainstream. By combining several AI technologies with automation and APIs I was able to produce a good number of pieces that would look just fabulous in a printed, coffee table art book. So I made one, and it's beautiful – but you won't find it in the bookstores. Let's explore why.

The world we live in is becoming more and more digital. The way we communicate, work, consume and learn has forever changed since the internet and the smartphone saw the light of day, and this shift was dramatically accelerated by Covid-19 over the last few years. Generative AI is now tightly linked to this new digital reality and, as a result, we have become increasingly surrounded by AI-generated images and videos in virtually every context imaginable. I was fortunate enough to experiment with this technology long before it became mainstream, and started to work with generative AI in 2021 already.

By now you will surely have played around with DALL-E, Midjourney or some other AI-based image generators yourself and you are probably aware of the level of detail and complexity displayed by their output. I will tell you, it's one thing to type in a quick prompt ("a blue cat playing piano") and get a neat image in a few seconds that looks kind of cool, like playing the lottery once – it's another thing entirely to combine several AI technologies, automation, and APIs to basically have:

• An AI companion helping you to come up with ideas for artwork.
• A language model drafting prompts with an extreme level of detail ("Oil-on-canvas painting of a blue night sky with roiling energy. A fuzzy and bright yellow crescent moon shining at the top. Below the exploding yellow stars and radiating swirls of blue, a distant village sits quietly on the right. Connecting earth and sky is a flame-like cypress tree with curling and swaying branches on the left. A church spire rises as a beacon over rolling blue hills.”)
• An automated process that runs this same prompt against generative AI many times, let's say 500 times.
• The Human curating the results, and selecting the very best.

That's why my results often stand out, and why some of the artwork I’ve produced looks so real that it’s hard to tell if it has been created by hand or by an AI. Of course, by now, the image generators I mentioned have become much more sophisticated and you can come up with complex artwork with a single prompt, but the story I want to tell you is not about the technology. It’s about how real world constraints bring a new dimension to AI-generated art.

By now, I have published two books: "Bedtime Stories from an AI" and "The Bot Who Loved Me", both exploring the boundaries between humans and AI. You can get the Kindle version of both, and they're also available in print. I loved designing the reading experience, the cover art, the lettering and typography – and to hold the physical books in my hands is a different, uniquely wonderful experience. Imagine, I sent you a link for downloading a free Kindle book, you'd say "Oh, nice, thanks!" but how much different would be if I handed you a physical copy of the book? Printing something in a book is not the same as putting this stuff on the web or making it available in a digital form.

The real world works differently. That's why I never published the book I had in mind that was to be filled with the artwork I was creating with generative AI. It's one thing to create a virtual art gallery with AI-generated images (that by now thousands of people have seen and enjoyed). It's another thing to print that same artwork in a book, and commit to selling it in stores and other platforms. The current discussions around copyright infringement and derivative works are not just theoretical – they are real and must be taken seriously. If the output of generative AI is substantially similar, or includes a substantial part of a third-party pre-existing work then, depending on its use, the user could also be liable for copyright infringement. This kind of artwork may be OK for a blog post or a tweet, but as soon as commercial or professional use comes into play, copyright holders will most likely look at the matter more critically.

So, although I now have enough material to publish a beautiful, coffee-table art book to be proud of, I'm not taking that final step. I do have a physical copy of it at home, and I'll probably pull it out of my bag when I meet you for coffee some day, but until the legal environment matures I won't be offering it for sale. And that's a story of its own.