The Effects and Ethicality of AI Art

The+Effects+and+Ethicality+of+AI+Art

What comes to a programmer’s mind when thinking of the phrase “AI art” may completely differ from an artist, TikTok user or simply another student at Cupertino High School. With the recent boom of computer-generated art being used on sites such as Dall-E, Tiktok and Midjourney, issues about ethicality, database biases and their effects on human artists have risen significantly, though opinions on these issues – and how to resolve them – vary widely.

One issue that many AI art filter users spotted was the bias in the generated images – often, a person with darker skin tone would be turned into something other than a person or whitewashed into a lighter skin tone. But because databases for these algorithms feed from the internet, which may be a biased source of pictures and art, programmers of similar AIs have shared that there is almost no way to remove these biases unless the model is trained on a less biased database.

However, the fact that these databases are pulled from the internet begs the question of if AI art should be, or can ethically be, used. The internet is full of art and training data, but those artists often never gave permission for this kind of use. Some have even custom-made their own algorithms based specifically on artists who did not give their consent or were completely unaware of the project. To make matters worse for these artists, it is legal to sell computer-generated art as long as the software allows it, which many do. Due to this, many artists have been repeatedly expressing their concerns over being replaced by an algorithm that uses work it had no permission to use. Thankfully, websites such as haveibeentrained.com can tell artists whether their work was used for training and if they have the rights to sue.

However, this still discourages many budding artists, especially ones who may not have the grounds to sue. Saranya Raghavendra, a junior at Cupertino High, is an aspiring professional artist that believes that AI art is harmful to artists. Said Raghavendra, “It’d be like if I made a collage [of others’ work] and claimed it was just me.”

One approach to attempting to resolve this issue is banning AI-generated images without watermarks. In December of 2022, China banned all AI-generated images that did not carry a watermark, following its 2019 ban of unmarked deepfakes. However, this does not ease the worries of many artists. Said Raghavendra, “Number one, a watermark is so easy to photoshop. Number two, it doesn’t help any artists. In my experience, people are still going to choose a watermarked [image] for cheaper over [a more expensive one] with no watermark.”

Another approach would be to ban AI art entirely. But that begins another question: What defines “art?” AI art began as computer-generated images of simple but unique ideas such as “cats eating lollipops,” which meant the model would need to train on images all over the internet. It is hard to dictate whether photography would count as an art form that needed permission to have pictures fed into the database, or whether the subject of a photo is the one who needs to give that permission.

With AI art filters and easily accessible programs, computer-generated art encapsulated the world in a short amount of time. But with all these issues in mind, maybe AI art is not all as great as it seems. Said Raghavendra, “Human creativity above all else.”