A blog related to original AI artworks. I make AI artworks using coined words and emoji as prompts.

Exhibition Results of “Art AI’s Free Creation Works” at Design Festa

I exhibited AI art pieces under the title “Art AI’s Free Creation Works” at Design Festa held at Tokyo Big Sight from November 11th to 12th. Although the AI art images on the MWAIP website showcase around 46,000 images, due to limited exhibition space, I displayed only 36 images. While most people unfamiliar with AI art walked past without noticing, many who were aware stopped by the booth and engaged in discussions. I also conversed with several painters, art-related individuals, and a person involved in cultural heritage. Particularly, I received sharp comments from some painters, and the reactions of others were informative. As I only sold postcards for 50 yen each on the day, the sales were minimal. However, some people purchased posters from Rakuten.

The title “Art AI’s Free Creation Works” derives from the idea of allowing AI to create artworks freely, using a minimal number of words (1-3 words) to prompt the creation of images. I only prepared 36 images and minimal descriptions for the exhibition. However, I feel that it would have been better to write several episodes to attract customers. I intend to do so for the next exhibition. As I couldn’t redesign the posters this time, I handwrote two additional episodes on the second day.

• When I input “Vermeer cat” into AI, it created a cat painting in the style of Vermeer. (Regarding the next painting)
  
  
  
• When I asked AI to “paint a painting in the style of Klimt, but not in the style of Klimt,” this is the resulting painting. (Regarding the next painting)
  

I received the following comments from painters and art-related individuals:

• “A painting in the style of Kandinsky resembles Kandinsky, but there’s no theme in the Klimt-style painting.”
  
  
  
• “Paintings created by AI are uniform and flat. The overall presentation feels tense.” I thought this was a sharp observation. People spend time drawing each part with tools, whereas AI draws everything at once, possibly leading to this outcome.

More than four kindergarten-aged children accompanied by their parents were captivated by exhibited paintings. They mentioned, “It looks like a face” (regarding the next painting) or “Amazing painting” (unable to specify which painting).

While I could only speak a little with foreigners, some merely said “interesting” and continued walking.

Other Notes

• “Previously, I adjusted the contrast when using Stable Diffusion.”
• DALLE 3 has limitations like not being able to input proper nouns.

It seems starting the conversation about AI art from ChatGPT was effective for those unfamiliar with AI art. Explaining that ChatGPT responds with text while AI art responds with images helped in understanding. Subsequently explaining AI art seemed to help them comprehend better.

First-day scene

Second-day scene