Since everyone is nonstop posting about their AI art explorations, I decided to join the party. I got invited for a beta test from Midjourney, and I thought if I’m using an AI to create whatever I want, why not let the AI create itself?

We’ve seen how we as humans, artists, filmmakers, designers, and creatives imagine what an AI looks like. Why not let the AI decide for itself this time.

To my disappointment, I was expecting to see something really whacky like the AI imagining itself to be a muscular dugong, a gigantic goldfish with a human brain, or even a flying hippopotamus. It’s interesting to see how consistent the AI imagines what ‘he’ looks like. Man in mid-’40s, glasses and humanoid feature + little robotic elements. It’s also weird that after generating more than 100+ portraits, I only get white male portraits but no women and other races at all. I guess the developer of this AI needs to focus a little more on DEI?

My takeaway:

First off, I didn’t create these, the AI did. What I did was type in a couple of keywords and the AI did the rest. I needed to clean these up with Photoshop extensively, I didn’t use Houdini or Octane Render to make it look realistic, I didn’t pick up my pencil or stylus and drew this, I didn’t create this ‘artwork’ from scratch, what I did was really log in to discord, type in a couple of keywords, watched it come back to me on Discord, saved it, cleaned up on photoshop and saved it. That’s a fact.

It’s still a nice tool for ideation and providing very specific direction from abstract thoughts, but none of these frames are ready to ship to a client. It’s amazing to have access to a tool that allows us to quickly illustrate an idea and share it during the pre-visualization stages of a project. As the technology gets mature, maybe years or decades later, I think it’s gonna somewhat make an impact in the creative industry.