The ethical dilemma of AI art
Over the past week, social media feeds saw a flurry of whimsical AI portraits proliferate.
Why are we seeing renderings of our friends, colleagues, and family members as fairies, elves, and the like? Because of the photo and video editing app, Lensa AI, just dropped a new ‘magic avatar’ feature that allows users to feed images of themselves into an AI program that uses Stable Diffusion to generate portraiture that appears as if created by digital artists.
All in good fun, right? Well, as it turns out, that might not always be the case. As popularity of the trend has skyrocketed, important conversations have re-emerged into mainstream consciousness regarding AI as it relates to data security, bias, and creative copyright.
To put it briefly, the use of AI for artistic purposes remains as divisive as ever and there are serious considerations that should be taken into account when considering utilizing AI.
AI is not perfect. It reflects the biases found in the datasets it pulls from, all of which are fed to the program by a human (read: biased) source. Every AI is trained off a given set of information. It accepts that given data as true and then learns to predict outcomes based on it. If the dataset reflects implicit bias, so too will the algorithm.
Some users have called out the Lensa app for being inherently fat-phobic and slimming down larger-bodied people — something that can be triggering for people with a history of body dysmorphia. Other users have voiced concerns over the apps perpetuation of the male gaze, with hyper-sexualized and feminized illustrations of women-identifying people being commonly generated.
In the artistic community, however, the predominant issue being discussed is that of AI generated art and its potential disregard of artistic consent and creative copyright.
Lensa feeds its AI algorithm from a database of images. Some artists have claimed that a number of images being fed into Lensa are actually copyrighted and are being used without the consent of the artists. This issue isn’t a new one, just days after the death of renowned South Korean Illustrator, Kim Jung Gi, a French game developer fed his images into an AI program and began generating images in the late artist’s signature style. Some people saw the move as a beautiful tribute to the artist, others saw it as exploitative. The artist did not consent to his artworks being used as datasets for AI and now some random person was profiting off of it — how could that possibly be ethical? The situation with Kim Jung Gi’s artwork being fed into AI datasets is not a one off, there are a number of living artists who are experiencing the same thing — their art is being used without their consent. When this happens, random people on the internet can very easily create believable knock-offs in the artist’s signature style, thus potentially damaging the artist’s livelihood.
The rise of AI has definitely been a tumultuous journey to follow, one that is by no means over. Up to this point, AI has been largely unregulated. Legal systems just can’t keep up with the speed at which technology is developing. If AI continues to be fed datasets that include artworks without the explicit consent of the artists, lawsuits and regulations will undoubtedly follow. Now, we wait and see what happens next.
