By Ben Yuri Biersach ’87

In one striking image, Peter Pan gazes out over Neverland, illuminated by the light of a crescent moon. In another, an 18th-century schooner in search of treasure drops anchor in the azure harbor of a tropical island. In a third, Alice tumbles down a surreal rabbit hole past jars levitating from a bookshelf. The visual craftsmanship in these and hundreds of other literature-inspired images is highly detailed, stylistically diverse and bursting with energy. All were created by Punahou eighth graders using generative artificial intelligence.

Case Middle School art department faculty member Hugh Mosher had been monitoring the evolution of AI’s image-making potential for several years and in 2024 felt the technology was advanced enough to finally introduce in class. “We need to include AI in our educational practice because it’s going to be a part of our students’ lives,” he says. “We have to give them some understanding of what the possibilities are.”

For the 2024 – 2025 school year, Mosher assigned a special project, requiring students to create a book cover with the use of AI. He asked Kosasa Learning Commons Librarian Lori Nakamoto ’04 Jones to select descriptive passages from five popular public domain children’s books: “Alice’s Adventures in Wonderland”, “A Christmas Carol”, “Peter Pan”, “Treasure Island”, and “The Wonderful Wizard of Oz”. Each student then chose an excerpt from the book and created a wrap-around book cover.

To bring this project to life, the students accessed Adobe Creative Cloud’s suite of graphic design, video editing and photography software. The suite’s Firefly platform is used to generate the images, and then the graphics editor Photoshop is employed to stitch each image pair together in the spine of the book. Since Adobe trains its AI models on only image stock it owns and public domain material, Mosher was able to impart an important lesson to his classes about the ethical and financial issues of creating and using copyrighted images. 

Another bonus: the platform is secure. “Adobe Firefly AI only creates imagery from the company’s library of proprietary images, not from the World Wide Web,” he says. This ensures that the content is curated and suitable for students.

At the start of the process, each student simply cut and pasted the chosen book passage into Firefly’s prompt window. “The AI initially creates a set of images which are usually a mediocre match with their descriptive passage selection,” Mosher explains.

CMS Students in Hugh Mosher’s art class brought their imagination to life, creating vibrant book covers of classics, such as “Treasure Island” and “Alice’s Adventures in Wonderland” – with the help of AI.

Students then evaluate the initial images and refine their next request to produce new outputs closer to their desired vision. “They need to be clever and rephrase the prompts to focus the AI generation on specific aspects of the scene,” Mosher says. For example, they might emphasize that ‘the house sits under a very dark sky’ or clarify that ‘Alice appears smaller than the bookshelf.’ 

A new image is generated. Not exactly what the student wants? They refine again. The more descriptive the language in a student’s prompt, the better the result. This revision process is the crux of what Mosher wants his students to master.

“I hit a lot of dead ends along the way,” recalls Finn Greenwell ’29, who took the class in the spring of 2025 and created a dark, moody depiction of the protagonist in “Treasure Island” surveying the book’s titular landmass from the crow’s nest of a ship. The first time Greenwell prompted Firefly, it placed his human character inside an actual bird’s nest rather than atop a maritime lookout platform.

“That was kind of crazy,” he says with a laugh. In the end, Greenwell applied more than 20 revision prompts to get the images to his liking.

The artwork generated by Firefly combined with the students’ skill in assembling the images yielded remarkable results. Many of the covers look as polished as those you’d see in a bookstore. Although the quality of the illustrations is far beyond what most eighth graders could produce if they had been asked to draw them by hand, the images clearly reflect each student’s vision. 

Mosher celebrates how advances in technology can democratize creativity and make art tools more accessible. “Suddenly, anybody can be a creator – anybody can come up with their own idea and generate it,” he says. “It used to take years of experience to learn to be a skilled artist and then illustrate your creation. Now, almost anyone with a computer can illustrate their ideas with AI, and that is a huge and important change. It is rapidly redefining the future role of the artist and illustrator.”

For Mosher, the most exciting part is seeing students use these tools to bring their imaginations to life in ways they never could have before.

Main Article: Prompted: Intentional AI Literacy for a New Frontier

Related Article: The Art of Iterating