The System to Replace Us

I must confess I use Artificial Intelligence. Not to write, of course. I would never use AI to write anything for me. Not because I am too morally superior to let a machine do my work. Generally speaking, I’m lazy and love shortcuts. The truth is I refuse to use Artificial Intelligence to write because I’m too prideful. I am way too talented of a wordsmith to need AI’s help with writing anything.

Where I know I lack talent and need to use whatever resources are available to me is art. It is shameful how pathetic my skills are considering my heritage. My Uncle is a professional painter and sculptor. A couple times a week, I walk through the building on my University campus that has his work on display just to remind myself how cool-adjacent I am. Like her brother, my mother is a phenomenally talented painter, photographer, and all around artist. She has an eye for lighting and creates incredible work full of heart. I have always dreamed of writing a children’s storybook with her illustrations. Perhaps one day we will.

But somehow, I missed out on the art gene. So when I need to come up with an illustration for one of my stories, I turn to the dreaded AI. At first I thought, is this cheating? But I concluded I am not representing myself as an artist, I am a writer. The picture is inconsequential compared to the words. It is a hook to draw people in, but as long as I clearly indicate it is created with an AI generator I think I am representing my work honestly.

But as I have used art generators, I have come to see the glaring holes in the bid for technology to make human artists (and writers) obsolete. The shocking truth is that not everyone can create masterpieces with Artificial Intelligence. Using AI isn’t always that easy. There is a skill involved in getting the generator to understand what it is you want to create. I am a wordsmith, so it should be simple enough to type out the description of what I envision. I paint pictures with my words all the time to transport my readers to a world that exists only in my mind. If I can do that, it should be easy as pie (also not really that easy) to tell AI what I want and get exactly what I intend.

It’s not.

In fact, sometimes the results are so frustrating, I want to scream. Other times, they are downright comical. Sometimes, I just give up and say what the hell, it’s close enough. Frequently, I take the flawed picture it created and crop out the errors. Yet every once in a while, I get exactly what I was hoping for.

The art generator I use is called NightCafe Studio. I learned about it from another Vocal creator over a year ago. I'm pretty sure it was Mikeydred (I have learned a lot from subscribing to him and highly recommend it.) I thought it was worth a try since I am artistically challenged. It is free. There is an option to pay for extra credits or features, but I get a few credits every week and have way more in my bank than I have used so far. There are options in NightCafe to share my work and enter contests. But I’m not an artist, I’m a writer, so I don’t do either of those things. I simply download the pictures I want to use for my stories and then upload them when it’s time to publish.

Someone recently complimented the picture I used in a story and it had me thinking about the process I went through to get it. The first “draft” didn’t turn out quite as I envisioned. For some reason it started as Anime style. I must have clicked the wrong button. I’m not opposed to anime style, but the part that really threw me off was that her head looks like it is backwards. (Perhaps she’s been beaming with Spaceballs.) I tried another version, also anime style. It was a little better but not quite what I had in mind. That's when I decided I needed a different style and added a bit more direction. That led to a fairly decent option. Her running mascara looks more like blood, but overall I got close to what I desired. I almost used it, but decided to add in plates of food and crystal goblets. The final picture was pretty good. But if you look at it closely, this time her running mascara looks almost like a giant hair hanging off her eye. I had given the instruction that her shoulders had red welts which led AI to add red splotches all over her arms, blood on her dress, and face. A little more than I wanted, but the overall emotion of the picture fit well enough with the story so I went with it. (I also wasn't sure AI knows what a quilt should look like; the final picture looks more like a throw blanket.) But I was very pleased with the aspen grove and look on the woman’s face.

Among my other frustrating attempts include this series. I wanted a picture with a mom, dad, and 3 little girls. No matter how I typed in the instructions, it absolutely refused to give me more than four people. The other problem, I wanted it to be by candlelight because in the story the power is out. NightCafe couldn’t understand that concept and kept lighting the Christmas tree. The only time it managed to keep the tree unlit, one of the children became a little boy. Finally, I settled for a different one and cropped out the lit tree.

This picture was supposed to be of five kids. It started out anime style again with only one boy and one girl. My second attempt tripled the children in the picture. I ended up changing my story to include another child because that was easier. I was in a rush to get it done after work before the challenge deadline and didn't have time to keep adjusting the instructions.

For this story, I just wanted the man to be in the driver seat and the woman to be his passenger. At one point it just moved the wheel to the right side of the car so it was in front of the man, but it also took away everything else I asked for and the story takes place in America where cars generally have the driver on the left. In the end, I switched gears entirely and asked for a picture depicting a different scene from the story. But my favorite variation will always be the first attempt in which AI decided to put the couple inside the engine compartment instead of inside the cab of the Jeep. I laugh every time I see it.

In creating this picture I asked for the woman to hold her fingers above the flame of the match. Two of the early attempts were decent, just not quite what I wanted, and the other two I found truly bizarre. Finally I simplified my instructions and went with the picture of the woman reaching toward the fire.

I never got what I wanted in this picture. I asked over and over for the woman to have her eyes open and only the man to have the blankets. I finally went with the third picture, but I had to crop it because if you look closely, you can see that the man has a blanket over his waist that then becomes his pants.

Perhaps the most mind boggling error I received came from the simple instruction, “golden balloons in the shape of 100 against a blue background.” Not at all complicated, yet I still ended up with the word “DO” instead. A second attempt gave me the number 10. So I improvised. I copied it, spliced it together, and cropped it to get some semblance of what I wanted. Good enough for my purposes.

Sometimes I am surprised and get exactly what I was looking for the first time, like in the following images:

To be fair, I have barely scratched the surface of what NightCafe can do. I have only used it 38 times. I know there are plenty of features I have not tried yet and if I took the time to explore the program I would probably learn how to better get exactly what I want. But that's the point. Art created with Artificial Intelligence doesn't just happen. It takes effort. Effort I have no interest putting in.

This reminded me of a conversation/good-natured argument I had on Christmas with my brother-in-law. Unlike me, Andy is a very talented artist and earns his living utilizing that skill in graphic design. I was walking through the room when I overheard him mention using AI. I immediately stopped, feigned deep offense, and berated him for using “the enemy” of the creative mind. No worries, neither of us was actually upset about the conversation. Instead we had an eye-opening tête-à-tête. He argued all the reasons why a good designer uses Artificial Intelligence as a starting point to help with the tedious tasks to increase productivity. The artist is then freed up to use his time on the focal point of the work. In the end we likened AI to a thesaurus (my favorite dinosaur.) In the right hands it can be a tool to enhance the skill of the artist. In the wrong hands you might get something like this abomination:

I would argue the same is probably true for AI generated writing, but as I mentioned before, I am way too prideful to try that.

I liken the worry over AI to many advancements in technology. Every time there is something new it seems the population generally splits into two camps. The first embraces it whole heartedly, eager to stay up with the times and never questioning any possible fallout. The second decries the advancement as the beginning of the end, leading to the inevitable conclusion that machines will make humans obsolete. The thing is, I just don’t see that happening. When self-checkout stands became all the rage, my sister worked at a department store and told me I should never use a self-checkout because it is taking a job from a person. A decade and a half later she uses them all the time. We have both seen that it doesn’t take away the cashier’s job, it changes it, makes it more efficient. It speeds up the process for customers who are now overseen by the same cashier they used to wait in line for, while they scan their own items. Now the customer can get on with their day more quickly, making the store more appealing and therefore increasing business. But there still needs to be a cashier nearby to help with any issues and answer questions, and, when necessary, tackle shoplifters.

In the same manner, Artificial Intelligence, like all technology, is a tool, but it still needs a human to input direction. It needs the creativity of an actual human to come up with the idea and tell it what to do. Once it creates something, AI can’t tell if what it made is quality or garbage. Again, it needs a human to discern if it achieved the goal. It doesn’t know the difference between cupping a hand around a match and holding a cup with a match in it. But I do.

Artificial Intelligence can aid in creation. But it has one major flaw that I am not sure it will ever overcome in its bid to overthrow humanity. AI is artificial and thus does not possess creativity. That still belongs only to the gods and living, breathing human beings.