I thought I’d share some of my personal opinions about the rise of AI art, the debates around it, and how it’s coming to take all the jobs of artists. This focuses on the likes of dalle-2 and midjourney. Which I have used both extensively. As a developer, I can totally sympathise with that fear of the unknown (github copilot, anyone?). But I think all our jobs are safe. At least for now…This is a developer perspective on AI art.
Does the AI think about it’s art?
One thing common about the current AI art generators is that it can’t function without a human prompt. It can’t consider things about the world and decide to draw it. You need to tell it what to draw. Like a child, it lacks wisdom and real-time experience of it’s environment and needs guidance (prompts). It also lacks any understanding of the meaning behind the art that it creates. What does a particular look, smile or gesture mean? Can it be asked how it feels about a piece of art it creates? This is an important difference between a human and machine learned artist. It’s not necessarily that one is considered more valuable than the other. They just approach the creating of art in different ways.
Imperfect imperfections
AI art still has it’s flaws in it’s execution. If you really study the art it creates, you will usually find some problems. This is the case especially when it’s created a human form. Limbs might not look quite right, and it’s not got the hang of drawing the right amount of fingers yet. Further prompts could probably fix this, but you’re soon spending a lot more time in refining your instructions for the ai. Time perhaps saved with a real artist who knows anatomy. Either way, there are plenty of resources out there, mocking AI art fails.
A creative outlet for all
What AI art does do well is level the playing field for non-artists. This is true for someone like myself, who might otherwise need a lot of time to study to become an artist. I can use AI art to quickly create something based on an creative idea. What was once the domain of people with the technical ability, is now open to anyone with a creative spark of imagination. Like most things, it has lowered the bar of entry across the board. Similar to creating videos, music, websites. Once these were the domain of specialists, are now more available to those who want to jump onboard. No prerequisites needed.
What about accessibility?
One thing I have noticed that people aren’t talking about, is the benefits of AI art for those with accessibility needs. You might have someone physically unable to pick up a paintbrush, but is able to prompt AI art with something like speech to text. For accessibility, AI in general would be a fantastic thing for empowerment.
It should enhance, not replace
It’s more likely AI will become a co-worker, rather than a replacement. The question to ask is, will one adapt to that or quit? As with accessibility, AI will make things easier for all and possibly remove the technical barriers which might otherwise put people off.
What are you afraid of?
I can understand how people can feel when they’re concerned it can make their job redundant or devalued. But if you’re at a particular level I can’t see how that would really be an issue. As a Web developer, there’s a huge amount of low level, easy to use tools and services which arguably could render my skills unnecessary. Website building tools like wix have lowered the bar of entry for users, making it possible for anyone to create a website. But there will always be a requirement for those with wisdom and experience in any field of work. I wouldn’t consider my role to be devalued because of this, despite the existence of AI which builds websites in 30 seconds. This is an area where a developer perspective on AI art becomes more relevant, and hits closer to home.
Legal issues
We also have to consider the legal issues facing use of AI art, some arguing questionable use of learning methods. It’s hard to say how this will turn out. Hopefully the AI developers will be transparent about the methods they use to train their systems. I’m not sure that they are just scraping images to generate images via some kind of montage. But it will be interesting to see what happens.
Note: I wrote this post, not AI 😀