I’m a photographer and artist (well, sort of an artist), among other things. While I am strictly an amateur at both, I’ve been using Photoshop for something like 20 years for editing photos, sketching, cartooning, etc. And I have some concerns about what’s going on in the world of photography and art. Let me give you an example.
Look at the photo over there on the left. It’s a fairly generic looking and pleasant image of an attractive young woman. A woman who doesn’t actually exist. That image was created a few minutes ago using Photoshop Beta with Adobe’s generative AI technology. As you can see, generative AI has become very, very good. Scary good, really.
Well okay there are still some “issues”, as they say. Let me who you the entire image that Photoshop came up with when I told it to generate a photograph of two women in a coffee shop.

See the problem? The young lady must really, really like her coffee. Or look at this one, which is another image generated using the same text description.

Generative AI may be very, very good, but it isn’t actually all that smart.
Let me show you another image. This one was generated with the exact same text description but this time I told it to produce a black and white line drawing. This is what it came up with.

Not the best artwork in the world but perfectly acceptable for a simple drawing. I’ve seen a heck of a lot worst human created artwork. And better still, no weird glitches with cups or hands. I got this one when I tried generating another line drawing with the same text description.

I’m sure you see the problem there. Adobe’s AI, and in fact almost all of the ones I’ve experimented with, all seem to have trouble dealing with hands for some reason.
Now, did you notice something else interesting about those images? All four images depict young Black women even though I did not specify a particular ethnicity in the description I typed in. I found this rather curious so I continued. Eventually I generated 20 different images using the same text description. Of those 20 images, 10 depicted Black women, 3 depicted women of Middle Eastern ethnicity, 4 were Asian, one was Indian and 3 were caucasian.


I found this rather curious. The same was true, although to a lesser extent, when I generated images of men. Don’t get me wrong. I’m not upset. In fact I find this rather refreshing because often we’re bombarded with images of nothing but white people in a lot of the videos, advertising, stock photography, etc. that we’re forced to endure every day.
But sometimes the images are a bit – well a bit disturbing. I asked it to come up with a line drawing of a couple sitting in a pub having a beer and this is what it came up with.

Oh, dear… Yeah, seriously. This is what it thinks a young couple in a pub looks like.
Just for the heck of it I added a specification that it should generate an image of a white couple and this is what it came up with.

Oh, dear again. Why does she have a pepper grinder glued to her head? Why is he dipping spaghetti in his beer?
But it also comes up with some stuff that is surprisingly good. I told it to generate an image of a grouchy old farmer leaning on a split rail fence watching a herd of grazing cows and it came up with this.

Not bad, really. Well, if you don’t look at his left had too closely.
For the heck of it I switched from photography to painting and it came up with this from the same prompt.

I actually like that one, to be honest. I would print that out in large format, put it in one of those overpriced tourists shops up in Door County and someone would probably pay actual money for that.
Which brings me to what I really want to talk about, and that is how AI generated artwork is going to effect society in general. We are already starting to see fundamental changes.
First, we’re going to see the end of stock photo providers ike Getty. Which is not necessarily a bad thing. Why would anyone bother to license stock images from companies like Getty when with a few words of description and Photoshop or one of the other AIs you can generate a generic illustration for your story or article? Let’s say you’re writing an article about vacations and you want to use a nice beach scene to illustrate it. So why go to a stock image company and buy the rights to use one when you can do this below with a few words and Photoshop?

Took all of about 15 seconds to generate that image from a one sentence prompt.
Stock image companies aren’t the only things that are going to come to an end. Would Normal Rockwell have ever gotten a job doing magazine covers in today’s world? I doubt it. I think illustrators in general are already losing their jobs, their work being done by AI. I’m seeing scores of images illustrating articles, fiction, even news stories, that look suspiciously like they were done by AI.
AI is already making inroads into the world of computer programming. There’s a very good chance that AIs are also already making significant decisions about you health care, not actual real doctors. The customer support person you chatted with on that website? Quite possibly an AI.
You may not like it. I don’t. Isn’t going to matter in the sightest. That horse is already out of the barn. Corporations can and will use AIs to replace human beings every place they can.

