I am still fooling around with this AI-generated image stuff. I've asked Grok and Gemini platforms to interpret poetry. The results are mixed and usually require several prompts before getting something worthwhile.
As an example, I fed Gemini one of my recent poems:
Fatally Shod
I'm wearing an old pair of pink bunny slippers.
A friend of mine claims they were once Jack the Ripper’s.
I took the damned things because they were free,
and now my old feet are killing me.
This was the initial result:
I then prompted with, "Perhaps with an older man seated in a comfortable chair."
The result was:
Finally, I prompted, "The slippers should have soles."
The result was:
The AI responses are sometimes very strange--the missing soles for instance. Grok is particularly strange when it comes to subsequent prompts to modify an image it created. If asked to produce a modified image with a simple change, it usually creates a completely new and wildly different image. In the example above, you can see that Gemini behaves in a similar fashion. Rather than slap some soles onto the first image, Gemini generated a completely different image.
I asked Grok about permissible use of prompt-generated AI images. From what "he" said, it sounds like the use of such images is safe from a copyright standpoint for now. If I had to guess, I'd say there will be lots of legal wrangling down the road. My prediction is that the use of such images (and AI-generated text, for that matter) will be treated like a writer's output. I foresee paying for one-time use rights, first serial rights, all rights, etc., just the way freelance writers have been managed for the past hundred or so years.
I wonder if AI developments will eliminate the need for human illustrators and generate a demand for "prompters." I can imagine sitting in a cubicle all day prompting AI to get just the right output. I suppose there will be some need for trained graphic editors to get rid of all those extra AI fingers.
.
