Using AI to develop visual concepts

Making AI image generation a tool instead of a threat

UPDATE 20/02/23: For an update on my thinking around the use of AI image generation, go here.

Over the summer AI image generation was all anyone was talking about. There was a mix of excitement and intrigue with fear and even loathing. Today I’m writing about an example of using AI as an assistive tool, rather than as a replacement for something else.

In October I published a chapter of Tales from the Triverse featuring a predatory creature called a ‘dopur’. Entirely fictional, and created for that specific storyline. I had the chapter mapped out in terms of plot and theme, but didn’t have a firm idea of the creature itself. That’s when I thought I’d try using AI image generation for some inspiration.

To set the scene, you can read the chapter here:

#57 Pet shop: part 1

The challenge I set for myself was to incorporate AI image generation into my overall creative process, rather than using it as a lazy shortcut. I wanted to avoid that “ta-da!” approach to AI image generation, with the image presented as-is and claiming to be art. Instead, I wanted the visuals to inform the writing, and vice versa.

I was surprised and pleased with the results, and how this process aided my writing. While there remain significant and legitimate ethical issues around AI generated images, there’s also great potential for it as a tool.

Let’s get into it, then: