Horsin' around with AI image Generation to learn AI prompt Engineering.

Back in the 90s, I was on a very famous TV show... Not really, but I was an HTML webmaster. I learned to take baby steps in writing in HTML from online chat rooms. Before Teams, Skype, ICQ, and AIM, we had nothing but plain text and HTML to enrich our text communications. In these archaic chat rooms, I learned how to use HTML tags. I later used this experience to jump-start a freelance business as a webmaster.

Thirty years later, I see a similar opportunity for everyone to learn the basics of prompt engineering by learning to use AI image generation. AI chatbots have surprisingly low user retention. Someone recently pointed out that as an IT professional, I have the knowledge base to take to this like a fish jumping upstream. Sure, it isn't easy, but I know enough about it to make continual progress, and I am confident I will reach my goal.

This made me reflect on my early days online, where I knew nothing with only a little Commodore gaming experience in my distant past. Before I started chatting online, I had multiple international penpals that I had met who were foreign exchange students visiting my school. So the idea of being able to communicate instantly for free with these friends and make new ones from around the world gave me the motivation to invest the energy in learning to chat and later learn to use HTML to fit in with the cool people who could imbed GIFs, color, and style to their posts.

Today, in corporate America, Social Media image is everything; we want our posts and presentations to stand out, just as I wanted my chat posts to stand out. Just like then, we all have access to a new tool we can learn to use to stand out: AI image generation. With the potential of AI image generation, the future of digital content creation looks bright and promising.

AI image generation uses the same interface, rules, and more simplistic prompt engineering to create excellent results. I made every image I posted in these articles using AI image creation, including this one.

I used DallE3 to create this image. I started with the prompt, "Bojack Horseman is writing HTML on the set of 'Horsin Around,' he is frustrated, sunset." Unfortunately, this resulted in an image with the screen of the laptop on the wrong side,

so I updated my prompt to "Bojack Horseman is writing HTML on the set of 'Horsin Around,' he is frustrated, sunset, looking over his shoulder looking at a computer screen with his reflection." I then requested the GEN-ID of the image and asked to convert the image to 1920x1080 pixels for Linkedin; then I had the image above.

After months of trial and error with AI image generation, I have learned that you need to state the image's perspective, lighting, location, style, and subject. I also used this image as a great example of the limitations of AI image generation. You can see the text is poorly rendered and could not show the ' in Hors'in. One of my articles had the word 'intangible'; AI will not include that word in anything because it is literally intangible. It is the equivalent of asking a computer to divide by zero.

Creating your style with AI image generation should take a short time. In your life, you are going to have ample opportunities to practice this skill, and at the same time, you are going to learn the logic and nuance of prompt engineering. Eventually, you are going to start having thoughts creep in, "I bet I could get an AI to do this for me by prompting..." This process is not just about learning a new skill, but about unleashing your creativity and creating your very own awesome style!