Tag Archives: AI

Andrew

AI Assistant

Leave a comment
New AI Assistant Sidebar

As I was prepping Maggie’s graduation posts for publication, I noticed a new WordPress sidebar called the “AI Assistant.” (It actually comes with Jetpack, a set of add-on tools that help optimize WordPress. Basic tools are free; more advanced usage comes with subscription fees.)

It looks like I initially got 20 “requests” for free. Improving the title or getting feedback on the post take one request each. Generating a featured image costs 10 requests. The feedback is pretty basic, but not entirely useless. It’s a bit like an advanced Grammarly analysis where it checks spelling and grammar but also gives tips on tone and style.

The featured image tool seems like it will be a direct threat to stock art companies. I ran it on my previous post about Maggie’s graduation slideshow. I thought it might create a collage or something from the gallery of images on the page. Nope. Instead, it appears to have generated an image based on an analysis of the title and text. Here’s what it came up with:

AI-generated image for one of the posts about Maggie’s graduation.

Definitely not going to use it. Far too corporate for my style. But I could see someone else using it and could easily see it as a generic graphic on an article about graduation in a print magazine or on-line zine. I can’t imagine how this sort of thing won’t cost real people their jobs.

Andrew, Griffin, Photo

The State of AI Art

Leave a comment

AI art is controversial. Some would say the phrase is an oxymoron. Or it’s more like AI Art Theft. I follow these discussions with interest and think that there is a lot to be concerned about. With that said, I also have a MidJourney account that I’ve been using for the past six months to play with the technology. I mostly use it for role-playing game material, particularly for illustrations of the fictional characters that inhabit the milieu.

While preparing for a birthday game session for Griffin and some of his friends, I needed to whip up a bunch of modern-day characters for the adventure (wherein a mundane plane flight would go horribly awry). Here’s the cast, giving you a sense of what MidJourney can produce these days.

Carlos Quintana, helicopter tour pilot
Cindi Bankemper, physics teacher
Duke Remington, police detective
Greg Stamour, paramedic
Tesha Janusz,
circus coach
Whiskey Sour,
drag queen

Note that for each image, I gave a very simple prompt. Whiskey Sour’s prompt, for example, was “middle-aged drag queen.” I never specified an art style, so it’s interesting that four of them are photo-realistic while Cindi and Duke are stylized. (Ms. Bankemper’s prompt was “science teacher middle-aged red-head glasses.”)

In a comical turn, on my first attempt at Greg Stamour, I typed in this prompt, “hot surfer dude paramedic.” MidJourney always produces four possibilities for each prompt. You can then choose one, add variations, or modify your prompt. One of the first four pics for Greg was this one:

hot surfer dude paramedic

Note the three pairs of sunglasses! Apparently MidJourney’s algorithm associates copious sunglasses with hot surfer dudes. If you zoom in, you can also see that the logo on his shirt is nonsense squiggles. Text-like artifacts are a common issue with generative AI.

I don’t know how all the ethical considerations will shake out in the coming years, but from a tech-geek standpoint, it is fun technology to play with.