CAA using generative AI in architectural history class

In his early days as an architecture student, Arshia (Ash) Alizadeh Moghadam sometimes struggled with identifying certain architectural styles in images.

His recent project through the Office of Undergraduate Research’s Summer Undergraduate Research Fellowship program was aimed at making that challenge a little easier for younger architecture students.

Earlier this year, Moghadam successfully created an AI-generated image which illustrates four distinct architectural designs for his faculty adviser, College of Art and Architecture (CAA) Assistant Professor Lori Smithey, as part of her research on using AI as a teaching tool. She plans to use Moghadam’s composite images this fall to help her students improve their visual identification skills.

“I feel like you learn more from seeing multiple designs in one image,” said Moghadam. “Creating this image was like putting a puzzle together. The hope is that students can deconstruct the puzzle and identify the individual design elements.”

Intelligent design

Moghadam’s participation in the project started when Smithey shared an idea she had for using AI to blend two architectural styles together and enhance student engagement with architectural history. She asked him if he would be interested in collaborating with her over the summer.

“Traditionally, analyzing images in architecture history meant showing two slides next to each other and asking students to compare and contrast them,” said Smithey. “I thought this might be an interesting way to take a step into generative AI and see if studying a composite image would help students develop deeper analytical skills.”

After they researched the best ways to approach the project, Moghadam began his initial attempts to create a unique image from two styles of architecture. After deciding on images he wanted to combine, he scanned them and gave the AI platform prompts about how to combine them into one image.

He quickly realized, however, that no matter how specific he was with the prompts, the resulting images were underwhelming.

“Whenever I gave it just two images, the results ended up being mostly one design with just a few small differences,” Moghadam said. “My premise was that if students see a blended image, they’ll become more aware of the smaller details within the image, which will help them understand and identify the influences better. So I needed an image with more examples.”

After several attempts, he found better results using four photos to create his composite.

“When I started using four images, it produced more of what I hoped,” he said. “A structure that didn’t looked mashed together but showed equal representation of each photo and had very visible design features.”

Moghadam presented his research results, including the blended image, at the Idaho Conference on Undergraduate Research last July.

Putting AI in architecture

Moghadam will serve as a teaching assistant in Smithey’s Global History of Architecture class this semester and may also discuss his project as a guest in other classes.

Later this fall, Smithey will present her research project results, which will include Moghadam’s work, at the Association of Collegiate Schools of Architecture’s research conference on AI design practices. The conference will feature discussions about the role of AI in architecture.

In the meantime, Smithey and other CAA faculty members will continue to find the best ways to integrate AI into their classes.

“There are some areas of architecture where AI likely won’t do a great job because the ideas are too complex,” Smithey said. “But I think Ash’s project is unique in that we’re using it as a tool to discuss history and help students develop their skill set.”

She plans to integrate some of Moghadam’s composite images into her course this fall — not to replace traditional slides, but to supplement them. Used as discussion prompts, the images offer a new kind of engagement — one that encourages students to analyze rather than memorize.

“In a class where we are lecturing, students are sitting back and receiving information,” said Smithey. “When Ash and I discussed the images he created, we weren’t just identifying features, we were having a conversation about what we saw. It opened up the possibility of using these composites not just to teach architectural vocabulary, but to engage students in visual analysis, interpretation and critique. That’s central to developing visual literacy.”

Moghadam’s work also helped Smithey imagine future assignments where students might generate their own AI composites, deepening their understanding through creative experimentation.

“AI is rapidly developing, and we’re still testing out its place in the field — what it is and isn’t good for,” she said. “But the results really depend on what the student brings to the project. They are the creative, human, critically informed part of architecture.”