Ellen M. Bassett, John Portman Chair and Dean of the Georgia Tech College of Design, center, with Gabe McGuire and Jemma Siegel who are explaining the process behind their project “Alglowrithm” in the Mason lobby on April 23, 2025. Alison Lumpkin (not pictured) was also part of the project team.
Mingling in the lobby of the Mason Building, a group of engineering students participated in their first art show opening on April 23. Like the hybrid class from which the artwork sprang, the exhibits were a cross between art festival and Capstone projects: drawings, paintings and videos accompanied by detailed informational posters describing the computational processes by which the collaborative final art was made.
Their artwork, created in collaboration with generative artificial intelligence (AI), was the culmination of a new interdisciplinary course called Art and Generative AI, co-taught by Francesco Fedele, associate professor of civil engineering, and Mark Leibert, a professor of the practice in the School of Literature, Media, and Communication.
Dr. Francesco Fedele
Professor Mark Liebert
Fedele and Leibert collaborated on the design of the course in response to call for more AI-based classes in the College of Engineering. “Art and Generative AI” is one of the electives in Georgia Tech’s new minor in Artificial Intelligence and Machine Learning, offered in partnership with the College of Engineering and the Ivan Allen College of Liberal Arts.
Taught for the first time in the spring semester, the course explores the intersection of AI technology, art, design, and neuroscience to create innovative synthetic, or generative media for artistic expression. Students learned how to leverage AI algorithms, design principles, and neuroscience insights to generate new forms of visual and auditory art.
Two histories, one goal
The students started out the semester getting a crash course in both art history and the history of AI.
Fedele he became interested in AI for art and rogue wave research. “Most people, when they want to learn something, they take a class. Me? When I want to learn something—I teach it,” he said.
He explained that he dove into the fundamentals, connecting them across disciplines, and built things—from neural models of ocean waves to brainwave-powered art installations. “I teach AI not just to share knowledge, but because teaching forces me to understand it deeply, ask better questions, and keep exploring.”
“The beginning is the perceptron,” Fedele said, referring to the fundamental model in machine learning, originally introduced by Warren McCulloch and Walter Pitts in 1943 that serves as the foundation for modern artificial neural networks. Fedele takes his students through the history of AI models and provides them with Python scripts to explore how each one works. Throughout the semester, the class inputs imagery into the algorithms to create composite art they can build upon.
“Basically, we deliberately push all these AI algorithms in their state of weakness so that they produce hallucinating and confusing output, Fedele said. “That is helpful, because then a student can reclaim control on the creative process and so they can use that output and reimagine it and reinterpret it.”
In contrast with the perfect output of modern-day AI software like ChatGPT, Fedele shows the students how to produce imperfect “hallucinating” output by deliberately limiting data sets, reducing training times or working with constrained models. The students then step in to interpret the imperfect output, elaborating upon it to reach the final artistic work.
Concurrently, the students also learned fundamentals of various artistic media — drawing, printmaking, oil painting and sculpture—in labs with co-teacher Liebert, a professional artist and professor of the practice who pioneered utilizing AI in artistic practice at Georgia Tech.
The students had a wide range of familiarity with artmaking, according to Leibert. “It's not like there are these really clean distinctions between humanities and engineering students,” Leibert said, “But it's always interesting to see how to make inroads to a unique student.”
Leibert saw the labs as working in parallel with the computer history, starting out with black and white charcoal drawings. “I would try to align that with the basics of computer vision, for example, binary images, black and white images. Then the progression of black and white to grayscale; the progression of grayscale to color, and so on,” he said.
Leibert stressed that as a teacher he always likes to “meet students where they are” in their creative process. “There are people who might think ‘Oh, I don't do art.’ But, it turns out they are really talented with manga stuff. You have to start with the things that people are good with and build the conversation from there.”
Leibert praised Fedele’s use of the early neural networks as examples for students learning AI. “Very few students have had access in any depth regarding the history of neural networks. This course is extremely unique in that regard and, I would say, probably stands alone,” Leibert said. “I've never heard of a course that covers the art historical depth and the technological depth that that this course covered,” he said.
CEE student Abigail Willis stands next to a still life and a self-portrait painted during Art and Generative AI’s lab class. (Photo by Michael Hunter)
The final artwork of the group, “A Blast from the Past,” Abigail Willis, Ryan Sage and Manasi Rathore, in collabaration with Generative AI. “An abstract painting that pulls feelings of nostalgia through colors, shapes and the faded feeling of the figures,” was the intended goal.
According to Fedele, co-teaching with an artist is a compelling model for integrating the arts into STEM, fostering creativity and innovation at the intersection of art, science and technology.
“I want the student to be exposed to different energies and ways of thinking,” Fedele explained.
Fedele became interested in art and drawing after a serious injury in 2016. He started drawing and painting during recovery as a form of self-taught art therapy. “As art became part of my life, I felt compelled to share its impact with my engineering students—helping them see science through an artist’s lens,” Fedele said.
This class builds on an earlier graduate-level class, “Arts and Geometry” that Fedele has taught five times since 2018. The class explores how geometry influenced the art of Pablo Picasso and the relativity theories of Albert Einstein. That course was also co-taught by Atlanta-based professional artists and culminated in the students showing their geometry-influenced artwork for one night at a local gallery.
Fedele was recognized in 2024 by Georgia Tech’s Center for Teaching and Learning with its Curriculum Innovation Award, which highlights faculty who are improving the quality of education at Georgia Tech through pedagogical and curricular innovation.
Empowering students with creativity
At the Mason lobby art show, the students displayed self-portraits and still lifes created during the artistic practice labs as well as their finished art piece created in collaboration with the generative AI. Projects used a wide variety of themes interpreted by the AI algorithms as the basis to build new art upon, including nostalgic imagery, Atlanta neighborhoods, contemporary light painting and the artwork of Henri Matisse. Each project explained the processes used to reach the final artwork created by the hands of the students.
The inaugural class proved to be a popular one, with 30 students from across disciplines.
“We had not just civil engineering students but LMC, Computer Science, ECE, Aerospace, Psychology, Physics,” Fedele said. “It's pretty broad-banded, which I like. Nowadays, engineering is so multi-disciplinary, you need the students to be exposed to different fields.”
Fedele hopes the class will equip his students with knowledge about AI’s true capabilities and limitations and prepare them to use these tools ethically and creatively.
“I think they will have a better understanding of what the fundamentals are when they utilize all these available software like ChatGPT, Mistral code or many others,” Fedele said. “They can approach it with more confidence.”
“AI entered our lives so rapidly, that many of us don’t fully understand how it works, why it works, when it fails, or what its core engine is,” Fedele said.
As for the future of AI, Fedele sees its true power as not in its ability to generate new content, but in helping us understand the content we already have. He thinks AI will revolutionize medical diagnostics by analyzing vast amounts of medical data to identify subtle patterns that might elude even skilled clinicians.
In Fedele’s own field of coastal and ocean engineering, AI can aid in predicting extreme events like storms, rogue waves, tsunamis and earthquakes.
To that end, Fedele, with the help of his class TA Kenneth Jenkins, has made his course notes on AI fundamentals available on a GitHub site for others’ use.
“The Python codes for use to make art with AI can also be used for research to forecast climate extremes, rogue waves, or to learn features of turbulence,” he said.
Fedele’s rogue wave research has been published internationally and he was invited to the the G7 at their third High Level Meeting on Maritime Security in 2017. His expertise was called upon in solving the 2015 sinking of the El Faro cargo ship during Hurricane Joaquin by the National Transportation Safety Board (NTSB). He asserts that he is using the same AI algorithms he taught students for making art to study the physics of rogue waves.
Fedele has also championed AI on campus during the spring semester. He moderated a public panel discussion called, “Being in the World — Will AI Ever Have a Soul?” that featured MINERVA, an AI agent contributing to the discussion along with panelists from Georgia Tech, Agnes Scott and the Goat Farm.
He, together with his students Kenneth Jenkins and Dennis Frank, also put on Improv AI, a dance performance on Koan Plaza on April 15, where dance teacher Amber Johnson and alumna Bekah Crosby from Dance Contempra improvised choreography to music generated by AI from brain waves. Georgia Tech Aerospace graduate student Daamini Visaalaakshi and Carol Subiño Sullivan, the assistant director of faculty teaching and learning initiatives for the Center for Teaching and Learning (CTL), also performed together with the Dance Contempra dancers.
“We collected the brain waves with an EEG headset and from the signals of the brain waves, they were converted into music,” Fedele explained. The dancers’ choice of movements was “human improv on top of the EEG.”
As always, human creativity is the focus for Fedele.
“It's not AI that rules the creation process, but it's the human that uses AI to create,” he said.