As far back as last August, there was much ado over the winning art entry for the Colorado State Fair. The winning piece of artwork wasn’t created by a human artist, but rather a piece of artificial intelligence software called Midjourney. With the help of a neural network and a huge picture library, this program takes written instructions and turns them into attractive graphics.
Artificial intelligence-based text-to-image generators have existed for some time, but until recently, their results were crude and lacking in detail. The work on display at the State Fair demonstrated a significant advancement in the complexity of this technology. Instantaneous, photorealistic picture production appeared, eliciting responses as intense as the artists’ original works.
Artists were mostly worried and critical, while tech fans praised the accomplishment. If anyone could create a painting in a matter of minutes, there would be no need to hire an artist or study art for years.
Another important concern brought up by artists is that the databases of these picture generators (DALL-E 2, Midjourney, Stable Diffusion) are mostly constructed upon existing photos from artists. The companies that made these tools took billions of photos from the internet, including works by artists who are still alive and those who have died. Artists may have trouble bringing legal action against generative AI developers due to fair use precedent, which these models presumably fall under, and the murky mechanism by which these computers choose certain pictures.
Some artists worry that this could eventually lead to them being replaced by computers that can make complex visuals on their own.
