By Lily Warner
Open AI is a research and deployment company based in San Francisco. It is home to some of the most advanced AGI (Artificial General Intelligence) research in the world; and has, over the past couple of years, been focused on text-to image generation. Open AI’s DALL-E came out in January of 2021, and showed the very real and present threat that AI represents to professional artists.
A text prompt, written by any individual, could generate a completely unique and visually pleasing piece of artwork. DALL-E could do this within a few minutes, for free. Unfortunately, it was limited in its ability to accurately respond to the prompt, and generated lower quality images when compared to professional artists.
DALL-E 2 changed that.
DALL-E 2 is a much more versatile and advanced version of DALL-E that was developed in April of 2022. It is able to generate images more accurately to the prompt and with 4x the resolution of DALL-E. If you write out “cow dancing on the sun during a sunset,” DALL-E 2 will create a selection of high-quality art pieces of that scene in just a few minutes.
Why, then, has it not already been used all over, for a cheap alternative to professional artists? The short answer is, it’s dangerous.
DALL-E 2 is capable of creating photo-realistic images that could be used in things like deep fakes. For example, it might create a realistic-looking photo of Tom Holland robbing a grocery store. Such images created by DALL-E 2 would be near-impossible to distinguish from legitimate photos. As a result, Open AI has kept its technology in limited access since its original creation while they developed security protocols.
However, a beta for DALL-E 2 has dropped recently, and the threat to professional art has suddenly become much more real. Open AI starts you off with a free 50 credits, with one credit equaling one use of DALL-E 2. Every subsequent month, you get another 15 free credits, and if you ever need more, you can pay $15 for another 115 uses of the technology.
All images generated by DALL-E 2 are fully owned by whoever inputted the prompt, and can be used for whatever purpose they wish. Up to and including commercialization of that artwork. Which, of course, represents an issue for professional artists.
If you could choose between paying, potentially hundreds of dollars for a single piece of high quality artwork, and paying 15 dollars for 115 art pieces of similar quality, you’d choose the cheaper option. As such, it will become much harder to become a working artist to survive than it already is.
That is not to say DALL-E 2 is impossible to overcome. It is not a perfect program, and has issues with composition (where things go in the image) and sometimes drops elements of the prompt if there are too many. Despite its downsides, it is so cheap that artists will have to be more skilled and practiced to compete in the new climate.
Being a professional artist today requires thousands of hours of practice, and with DALL-E 2, artists will have to be better than ever. That requires more practice, more time, and far more resources. Artists are already notorious for being unable to pay off student debt, but with an even higher entrance level necessary to overcome DALL-E 2Professional artists will never disappear. There will always be some people able to scrape by making a living on their art, or lucky enough to be born in a situation where they can devote themselves to their art free of worry. However, DALL-E 2 will make the industry much smaller and far more difficult to join.