CURRENT POSITION Assistant Professor and Digital Humanities Librarian, University Library, University of Illinois Urbana-Champaign DEGREE Currently completing MSLIS at University of Illinois Urbana-Champaign FAST FACT Ton and her husband are blacksmiths who work predominately with recycled materials. FOLLOW news.illinois.edu/how-are-deepfakes-being-used; bit.ly/recognizing-excellence-NEH-grant; bit.ly/fighting-deepfakes-hallucinations Photo by Robb Cohen Photography Company

Machine Learning

Mary Ton thinks there’s an ethical place for AI in the arts and humanities. She teaches researchers and students about the benefits and limitations of AI in research and the arts, drawing on her background in machine learning, and has presented on the topic across the state. 

Ton understands why librarians are wary of AI, since the tools can generate fake information. “My approach has been to explain how AI works, identify its limitations, and create a welcoming environment for experimentation,” she notes. “Once we have that foundation, those guardrails, so to speak, we can imagine possible applications.” In addition to designing a popular workshop series about AI, she developed a module to help instructors set rules around AI use in the classroom. 

She warns users against sharing sensitive information with an AI tool and suggests double-checking information with outside sources. But she sees potential for AI in libraries, such as creating metadata or improving transcriptions of digitized materials, explaining, “These use cases aren’t intended to replace humans or to devalue human labor but to help us improve accessibility at scale.” 

That includes access to the library’s digital collections. As part of the “No Longer at the Margins” project, Ton is helping digitize materials from her library’s domestic science program to emphasize the contributions of women to science. “I’m very curious to see how large language models could potentially streamline the transcription process and how library special collections can introduce data scientists to humanities concepts,” she says.