ChatGPT 4o images puts the biggest robbery of intellectual property in history to the next level. Over the past days my LinkedIn feed is flooded with postings about 4o images. Most of them are excited about the „new fantastic opportunities“, about the „cheap access to design“, about „the quality of text rendering“, about how „everybody could create designs and images in certain styles“, etc. etc.
It’s two narratives that are repeated again and again:
- Now design is cheap and available for everyone
- AI is the magic wand that makes everybody a creator with unlimited possibilities
So, let’s have a look at these in more detail:
Design is cheap and available for everyone. The truth is, it is not cheap, it is just paid somewhere else. By the creative people whose work was and is used for training with any permission and without paying any compensation. I mean think about it: am I as a person allowed to grab images from various artists and use it for my work with at least the permission of the creator? No, I am not. And If I copy the style of an artist 1to1 I will be called a copycat and if I try to sell „my work“ I would get sued by the original artist (rightly).
The machine would not be able to create all these „beautiful“ images without the training data of real artists, designers or photographers. The output of the image generators is only possible because of this work. The value of ChatGPT and all the other Gen AI tools is based on the work of real designers, artists, writers, photographers, etc. So, value is created out of the intellectual property of others, without paying them even a cent.
And this is done on purpose by the big tech companies. Another example: Meta used millions of books, ebooks and research papers to train its Llama AI without permission, consent or compensation. On purpose they used LibGen, a pirated library that includes more than 7.5 million books and 80 million research papers.
Big tech is stealing and using intellectual property to make money. End of story. And when we use these tools, we make a statement that we think this is ok. So let’s ask ourselves: do we really think this is ok?
—
And with this let’s have a look at the second narrative: AI is the magic wand that makes us all super creative with unlimited possibilities. Let’s just think about: how creative is copying someone else? And even if we think that is ok and creative. What do you think, how many other people create very similar images? Because what Gen AI does in the end, it sets conformity. The outputs become more and more similar. This is because of the nature of Gen AI. The strongest signals in datasets will prevail, used again and again and by this even get stronger. Weak signals will vanish.
What is true creativity?
But what is true creativity? Substantially new things always occur from doing something totally different from how it has been done before. Gen AI cannot do this. Gen AI will always recreate what it already knows. It will never do things differently. Remixing is also a form of creativity and Ben AI is good at that. It’s not creating something really new. We should think about that the next time we use it to „create“.
And that brings me to the last point: by using Gen AI for example by uploading our images to it to create „something new“ or „something in a new style“, we are giving our own intellectual property, my image, my face, etc. to the machine. Open AI will keep every image you upload to it. And for sure they will use this data for future training.
So, in the end it’s not only the artist, designer, writers, photographers they are stealing from. It’s also us, you, me, everybody using it.
Thorsten Jonas is the Founder of the SUX Network and a Sustainable UX and Responsible AI consultant and keynote speaker. He is guiding teams and companies in crafting sustainable, responsible and ethical digital products and co-leads as invited expert the UX Chapter of the W3C Web Sustainability Guidelines.
