Last week I fell into AI image generation. I signed up for the MidJourney beta and was immediately obsessed.

I have way more vision than artistic skill, so plugging insane concepts into a robot suits me down to the ground. You then refine and refine until you arrive at your perfect nightmare.

My bubblegum circus series:

My memento mori series:

My storm series (or Why We Do Not Trust the Sea OK)

It’s not just me

Most AI art tends towards darkness. I think some of that is us humans giving it dark concepts because that’s what we’ve come to expect – it’s futuristic dystopia to us, so that’s what we feed it.

However, I also think that we interpret the output as dark. The first images you’re given are often disordered and confusing, which our brains don’t like at all. We then try to refine them into something more recognisable, but we’ve started with oddness. Anything we build from there is likely to be disturbing.

AI does well with dreamscape-style stuff, too. There is lightness to be found but, more often than not, I’m seeing nightmares.