Does C-AI ‘steal’?

I think the answer is ‘yes’ but perhaps in a different way than you might think. It’s complicated.

Picasso famously said “Good artists copy. Great artists steal,” by which he acknowledged that all artists stand on the shoulders of those before them and that every piece of art is a remix of what came before. Yes, AI remixes similarly to how human artists do.

However, can remixing of artwork be considered ‘fair use’ when ‘the artist’ is a server rack in a high security data center of a tech giant that bases its ‘art’ on all online available art pieces?

Some people argue: C-AI ‘takes inspiration’ from other artists, just as living artists do. while some argue that C-AI, like the iron machines of the Industrial Revolution, is simply another part of the unstoppable progress of technology, so it’s best to ‘deal with it and adapt’.

— I agree and disagree with both arguments, in different ways. So many things AI are paradoxical.

Is AI truly ‘creative’?

Yes, I would argue that creativity (the ability to bring into existence a new solution, method or artistic object or form) is what C-AI has proven to posses in a variety of ways.

Like every human artist has been influenced by the works of other artists, so does C-AI ‘create’ new images by recombining the influences of prior work. And because LLM’s are part of the visual AI, C-AI actually ‘understands’ the real world relationships between different concepts, objects and tensions in an image.

While it seems like ‘common sense’ that computer algorithms can’t be truly creative, I invite you to reconsider this.

• First of all, neural networks in AI-systems mimic the organic structuring of the human brain and in that way C-AI behaves far more freely and randomly associative than we are used to from computers.

• Secondly, the ‘diffusion’ part in most AI systems adds an element of unpredictability to each output, resulting in unique results each time.

Organic-like structuring of data combined with diffusion and other AI data processing techniques lead to a certain level of synthetic-artistic ‘intuition’ that enable C-AI’s creative capabilities.

Was C-AI ‘just inspired’ by prior work?

No, C-AI does not ‘get inspired’ by billions of images. Inspiration (the process of being mentally stimulated to do or feel something, especially to do something creative) remains exclusively a human thing. An AI system isn’t mentally stimulated in any way as so far, there is no evidence of consciousness or emotion in AI systems.

AI ‘understands’ the world like a person would understand a delicious soup that was only explained to them but not tasted. When C-AI creates beautiful work, it’s only able to do so because it channels the heart, soul and inspiration of human-created work.

C-AI knows how to cook the soup, but doesn’t know its tastes.

Does C-AI ‘Photoshop’ prior art?

It doesn't, because technically, C-AI systems don't ‘cut and paste’. Instead, they create truly unique new artworks based on the design patterns and principles of the artworks in the datasets, that are embedded in neural networks.

This raises a new question: can design patterns and principles be stolen? The legal answer is probably ‘no’, because how would you as a maker technically argue to a judge that your ‘design patterns and principles were stolen’?

You couldn’t.

Does AI steal specific artworks?

Technically, I don’t think that C-AI steals particular artworks in most cases. It only does so when a particular artwork is the only one associated with a certain word in the learning dataset, a problem that is likely to be solved by adding more diffusion to the mix on data poor prompt words.

What I believe that C-AI steals is the inspiration, soulfulness and personality of human made art that are embedded in design patterns and principles. This type of theft might actually be far worse than copying artworks, because it means that anyone can output ‘unique’ works in the style of any particular artist or a combination of them.

C-AI will mimic humans artists better over time as the neural networks keep learning.

While not of the quality as physical paintings by the artist, do these images channel the spirit of Dali?

Should we ‘just adapt’ to C-AI?

Yes — it’s true that as people we must always adapt to changing circumstances such as advancing technology. But like workers living at the start of the Industrial Revolution, do we really have a choice? While there are lots of similarities between the impacts of Industrial Revolution machinery on workers and the systems of C-AI on creative professionals now, I think there is also one distinct difference.

While the invention of the iron machines was purely the result of science and engineering, the enormous datasets that empower visual C-AI, such as LAION, owe their enormous creative power entirely and exclusively to all the artists whose work has been harvested by web crawling bots online.

The machines of Creative AI now selling ‘artworks’ for a few cents apiece, could not have existed at all without having been ‘fed’, ‘loaded’, ‘supplied’ or ‘powered’ directly by generations of original and inspired, painstakingly created human art from both dead and living artists who would often dedicate their heart and soul to their art their entire lives.

It seems very clear to me that Big Tech owes the entire arts and creative sector, not just a little bit, but Big Time and this is why collective action is urgent and essential to secure the future of artists and designers.