This may the longest title in our database. In fact, it’s a set of instructions—cleaned up for clarity—given to an artificial intelligence (AI) which generates images based on a verbal description. This image was submitted by DS from Germany who wanted to begin a debate regarding the ethics and capabilities of emerging AI.
Stable Diffusion AI et al – Beautiful 8-year-old girl topless in profile, gold applications, floral patterns, by Gustav Klimt and Alphonse Mucha
This particular image was created using a free application called Stable Diffusion which, like many in its cohort, create images from text. This is an early stage of developing AI technologies that simulate human pattern recognition and generation by “training” a computer on a multutide of examples on the internet or other massive database. After “seeing” endless examples of various objects and concepts, the computer is finally able to recognize a car or a chicken or a shoe, etc. which can appear in a myriad of variations/styles.
The technology is far from perfect so some quality adjustments need to be made. For instance, DS usually generates hundreds of images on his computer (about two images per minute) and selects the best ones. The algorithm in this case is not error-free, often producing too many fingers or heads appear. By repeating the process, an acceptable form can be found or some reworking may be required using Photoshop.
The more professional versions charge a fee but there are some free versions besides Stable Diffusion such as Midjourney and OpenAI.
There are many ethical issues that come to mind using this technology, some more obvious than others. And it is important we begin to deal with these sooner rather than later. After all, it is not as if we can put the genie back in the bottle. What control do we have over our own image? What about professional models whose images are used? What compansation is due to living artists whose work is used to train the algorithm? What is the potential for fraud in the art collecting community? What freedoms or limits do there need to be with respect to standards of decency?
These are just the tip of the iceberg and the last point forces us to expound on the issue regarding minors portrayed engaging in illegal activity. Already, many companies have put in safeguards to prohibit nudity, but right now it is a simple matter for competent hackers to override this safeguard. Let the discussion begin!
when photography was invented, painters thought that was the end for them. I think we are in the same situation, the world is changing and we have to adapt. New technologies are invented, with their advantages and their problems, but there will always be artists capable of dazzling the world with their art that no machine can replace. AI can be interesting for an artist to find ideas, to explore tracks he wouldn’t have thought of, to find compositions and inspiration. On the other hand, the image database on which the AI is based is a problem in terms of copyright and should clearly be regulated. The theft of creation on the internet is not new, the problem here is that it is hidden behind an intelligence which will show only a result and never from what it was based to produce this result… Maybe a solution can be to force the developers to pop up with each result the different artistic sources that have been used and to have a reference of them (metadata) in all generated images.
It’s a very big debate but to resume, I think if it’s correctly regulated, for an artist it can be at the end a source of copyright incomes (like the sampling in the music industry)
The image is described in the title as topless and is tagged “nude”. Postcard dealers, and probably others, refer to all topless images as nude. This seems inacurrate even in reference to images of adults, and it seems especially strange to say that an image of a topless 8-year-old is a nude image. Does Pigtails have a rule that tells when an image is “nude” as opposed to “topless” or “seminude” or “dressed”?
We tend to err on the side of conservatism. Topless will often still be classified as “nude” somewhat depending on context. In most cases, full frontal nudity will be restricted to help mitigate shock value and can be viewed only by registered users. -Ron
I am in charge of categories and tags. We limit the number of available tags, so we have “Nude”, but neither “Semi-nude” nor “Topless”.
Discussion of intellectual property rights in the electronic age is all well and good, but I can’t be the only one who finds the sample image absolutely horrific. I can only hope the technology improves substantially, regardless of legal ramifications.
Indeed, we are at a beginning and there are weird side-effects on the way AI interprets the data it consumes. I think because this image was based on an artistic style, the image is not so bad considering. However, I have already received submissions from others who are creating “photorealistic” images. I assure these images are even creepier. For instance, one example I might share is busts of young girls but one of the things I noticed is that the AI does not understand the correlation between age and skin quality. The texture of the skin is realistic in a way but not for a child that age. I am thinking about maybe sharing one example per month because I expect to get inundated now that we opened this subject. -Ron
Stable Diffusion is built from the LAION-5B dataset, which is full of stolen copyrighted work. It was done under pretense of academic research, but now they are using it to raise investor money and paid subscriptions.
Also, the way the software is constructed, it is the easiest to have it produce a duplicate of a picture encoded in it – it takes special effort to make it NOT produce a duplicate, and even then it keeps doing that. The only reason that we cannot immediately see which original images it plagiarizes on each occasion is that the data set is huge: more than half a billion images in the basic set alone, which makes the chance that a human being familiar with the plagiarized work will see it in Stable Diffusion’s output is very small. But still there are such cases, and searching LAION-5B shows that it basically consists of work scraped off the Internet, violating a huge number of copyrights and license agreements.
Stable Diffusion was created on intellectual property theft, to begin with. They’ve put billions of man-hours of other people’s work into it which they had no right to.
Thank you. This is the kind of background information I think readers and the general public should know about. But, to be fair, some debate needs to take place about the dysfunctional nature of copyright law in the face of new technologies. -Ron
A human can observe 8-year -old girls and the works of Gustav Klimt and Alphonse Mucha. Then, using a paint brush and canvas, he can create an image of a girl in a style inspired by Klimt and Mucha. A human or group of humans can also write an AI program and cause a computer to make the image of a girl in a style inspired by Klimt and Mucha. In my opinion, any determination as to whether any compensation is due to artists or models, or whether the image is indecent, should be based solely on the image, without regard as to how it was created. If an artist wants to sue for copyright infringement, the jury need only see the image to decide if it is an infringement. They need not know if it was created with a paint brush or with AI and a computer.
You point about the irrelevance of medium is well-taken. But it is not simply a matter of medium, but on the material the algorithm is trained on. The lack of transparency is at issue because some of the material may be based on materials that are still copyrighted. Klimt and Mucha are certainly not at issue but it is yet another of a myriad of black boxes that will keep us from understanding what makes things work. The real legal issues may actually be trivial at this time, but this is yet another stress point for those who have to adapt to a rapidly changing world. -Ron