Current Thoughts on Use of AI Genreated Images

Posted: March 14, 2024

I felt myself getting a little heated in a conversation about AI and ethics this week (specifically images created by AI). I'm sure many others who work in media production are seeing the trickles of "can we just do this with AI?" begin to seep in.

Before I dive to deep let me clarify that I am in no way completely against AI or it's use. I've always considered myself an optimistic futurist to some extent; and I see a lot of positives using large learning models and other software tools in this vein. I use some already in some day to day for both work and personally. I do, however, think that we need to truly think the through the ethical ramifications of how it's used. We have to not just think of every problem it could solve, but think through what it's actually useful for and good at.

If using DALLĀ·E 3 or Midjourney was truly just giving a bot a prompt and it somehow divined an image from it's own mind then there really wouldn't be problem; but in reality these learning models are being fed with an insane amount of training data. Large amounts of which land in an ethical grey area at least for how it's obtained and how much power internet users have to not participate.

When we are using these stills or even soon videos; we are benefiting off regurgitating the hours and hours of hard work done by human artists. Unless we can find a more ethical way to source enough training data to make the results good enough I don't see how we can ethically use these in any professional setting. The tool is cool, but we have to stand up for the human artists without whom the tool wouldn't even work.