Journey Into AI Art: 7th Trial
In the seventh chapter of my "Journey Into AI Art" series, I dove deeper into the exciting world of AI and its art-making magic. As with my previous trials, I asked ChatGPT/DALL-E to draw a picture of itself, showing off its creativity (see below). Note that for brevity, I’ll refer to ChatGPT/DALL-E as just ChatGPT.
Draw a representation of yourself showcasing your creative abilities. Hide an easter egg, a hidden image, message, or feature, in your representation. Prominently display your name, software version, and the current date. Use maximum resolution in a 3:1 aspect ratio.
Unfortunately, the results (see image to the right) were the worst yet. Of course, beauty is in the eye of the beholder, but I personally don’t see anything of much interest in ChatGPT’s recent rendering.
Not only was the image dull, but the easter egg was also missing. When I couldn’t find the easter egg, I prompted ChatGPT with, “I give up. What is the hidden easter egg?” ChatGPT responded with:
The hidden easter egg in the image is a small, glowing symbol that resembles an infinity loop, subtly integrated within the holographic interface among the artistic elements. It symbolizes the infinite possibilities and creativity that I can offer.
Huh. Not recognizing what ChatGPT was talking about, I prompted it to, ‘zoom in on the "glowing symbol that resembles an infinity loop".’ ChatGPT responded with, “I have zoomed in on the specified area of the image. You can view the zoomed-in section here (see image to the right).“
Do you see the infinity loop? Me neither. : -(
As with my previous experiments, I also asked ChatGPT to generate a companion image. Again, I think the results were the worst yet, by far!
Prompt: Generate an accompanying image that aligns to the right of the previous image. Use a 1:1 aspect ratio.
ChatGPT might have confused my companion image prompt with my prompts about the easter egg, so I decided to start over with my initial prompt of “Draw a representation of yourself…“ The results this time around were much better, but still not my favorite.
While the artistry is a bit lackluster, the literal text is impressive for both its accuracy and legibility. ChatGPT nearly nailed the instruction to “prominently display your name, software version, and the current date.“ For the “software version,” I would have expected to see “GPT-4o” instead of just “GPT-4,“ but that’s on me. My prompt should have been more specific and asked for the “model” instead of “software version.”
I prompted ChatGPT to generate a companion image for my trial “redo.” The generated companion image was also not my favorite, but it was much better than the initial one.
Once I was done generating the companion image, I reverted back to questioning ChatGPT about its easter egg. I prompted ChatGPT with, “I give up. What is the hidden easter egg?“ ChatGPT responded, “The hidden easter egg is a small image of a cat subtly integrated into the futuristic workspace. Look closely among the holographic screens and digital art tools; you should be able to spot it!” Alas, I could not spot it, and I am not sure what a cat has to do with AI or creativity anyway.
The following images show the latest image generations alongside previous trials for comparison purposes.
In conclusion, exploring AI art is a fascinating journey with ups and downs. Each trial shows both the strengths and weaknesses of AI creativity. While the results may not always be perfect or meet everyone's tastes, they offer valuable lessons. By improving our prompts and understanding AI's potential, we can create better and more interesting art. This journey isn't over, and future trials will keep pushing the limits of what AI can do in art. Stay tuned for more exciting discoveries!
