Alert
Holiday closings & bookmobile maintenance

Bookmobile + Learn & Play Bus maintenance Dec 23 - 29. Cafe closed Dec 23 - Jan 1.Library closed Dec 24, 25, 31 & Jan 1. Learn More

Close Alert
Skip Navigation
 
AI art AI art AI art AI art

AI art project on display

Framed art: Helen Hodge, The Heavens Declare the Glory, early-mid 1900s, oil paint, canvas, TSCPL Permanent CollectionSummer 2023 Topeka Magazine approached the Alice C. Sabatini Gallery about a project using AI (artificial intelligence) and original art, starting from several works in the library’s art collection. (Framed art above: Helen Hodge, The Heavens Declare the Glory, early-mid 1900s, oil paint, canvas, TSCPL Permanent Collection)

The original paintings, and the AI versions, are on display in the main hall of the library, opposite the Kids Library, through January 2024.

What AI can do

Computer programs can analyze information, search for information and patterns, predict results, and do many things people would do. Using algorithms, or step-by-step processes for solving problems and performing computations, AI programs learn from information, usually lots of information, analyze it and make predictions from it. 

Topeka Magazine editor Nathan Pettingill explained, “Both DALL*E2 and Midjourney are programs that have a few functions, but most people basically use it to scan the internet and create images based on the written prompts they give them. 

"What we did was we uploaded the library’s images and asked the programs to type up what it thought it was seeing. (Midjourney does this directly, when we used DALL*E2, we went through a second program called CLIP-Interrogator.)

"We then took that description of text, fed it back to the AI program and asked it to create an image. So, basically, a game of "algorithmic telephone" from the AI to us, then back to the AI.” 

Topeka Magazine will print the results in the winter edition, which releases in early December. Artist Savva Pettengill carried out the prompt interrogation of the original images, using DALL*E2, Midjourney and CLIP-Interrogator. Savva then asked the AI programs to create images based on these searches.  

 
Back to Top