Human touch, digital mind
Richmond charts the course for digital humanities and AI
Meet Dr. Lauren Tilton, a digital humanities professor at the University of Richmond. When it comes to artificial intelligence, she knows a lot about how the algorithm and the artist meet. Her academic research focuses on feeding liberal arts values and thinking into AI technology, particularly as it relates to history and culture. Her projects have married computer programming with photography and film to create new platforms that help tell the story of the past.
“A lot of my work has been thinking about how to use AI methods in unexpected ways,” she says.
Tilton is leading the rollout of the new Center for Liberal Arts and AI (CLAAI) debuting at Richmond this fall. Her goal is to emphasize the need for interdisciplinary computational approaches in education and AI research. Students will benefit in a variety of ways that include advancements in AI technology, understanding how the arts and sciences factor into the technology, and learning how to stand out in the future labor market. Afterall, tech companies are already showing great interest in the intersection of AI and the humanities.
Tilton says she’s most excited about the students who specifically attend a liberal arts university, like Richmond, because they want to think differently about how to use AI in tandem with their love of art, English, literature and other disciplines.
“Digital humanities offer them that outlet,” Tilton explains. “So, you can be an art historian and do data science.”
Tilton is no stranger to blending history and AI. Fifteen years ago, while a graduate student at Yale, she co-led the creation of Photogrammar, a photography-based AI project. Tilton brought the project to UR when she joined the faculty in 2015 and partnered with UR’s Digital Scholarship Lab to launch the site.
Tilton and the research team applied computational methods to a large collection of historic American photography. The goal was to take rare photographs captured during the Great Depression and World War II and make them accessible to people around the world using AI. Before Photogrammar, the image collection was only housed in the Library of Congress, making it relatively inaccessible without visiting Washington, D.C.
Tilton says on the surface, the process of inputting 170,000 photos into an algorithm to create a web-based search tool seems straightforward. But capturing the nuances of each photograph was quite complex and challenging.
“The data of photography, the data of film, the data of historical record, [it’s] incredibly messy,” she says. “Trying to work with AI to garner insights from messy data is one of the hardest things you can do, and it takes a real liberal arts education to do that.”
Tilton says it all boils down to being able to ask good questions, and students need the right training ground to learn how to do it.
“My tech buddies say it like this: ‘ChatGPT can teach anyone how to program now, but it can’t teach someone how to ask a good question and garner insights from a bunch of complex information,” says Tilton. “And who does that better than the combination of the humanities with cutting-edge methods?’”
Currently, Tilton is studying television and film archives, giving special attention to how the storytelling techniques unfold and asking questions like: Why does the camera pan, tilt, or go in for a close-up? Why are certain actors in the foreground or background? Why is certain music selected? Who’s on screen or off screen?
Tilton says whether we realize it or not, as viewers, these visual signals and messages in film and television are shaping our values, ideas, and belief systems. She now wants AI to think about that. “The future right now is multimodality. To put all these elements together algorithmically,” she adds.
Tilton says she appreciates how Richmond is doing AI differently than other institutions and centering the humanities in the movement. She says Richmond takes the “unexpected approach” and that has been her reward.
“We don’t follow. We go the unexpected path all the time,” Tilton said. “We experiment, we go off road, we come back on road, we don’t just follow the crowd.”
A lot of my work has been thinking about how to use AI methods in unexpected ways.