Cornell’s AI Initiative supports university faculty, staff, and students as they explore models to optimize the use of generative AI for practical purposes. On January 22, 2025, eight local teams stunned a hybrid audience with presentations proving generative AI applications could simplify and solve problems that previously required intense manual effort. Each project team was led by one or two faculty or staff members and supported by students from the Cornell Ann S. Bowers College of Computing and Information Science (Cornell Bowers CIS).

“The students gain valuable experiences that put their AI knowledge to the test, and at the same time they are helping their university clients. It’s a win-win,” said Thorsten Joachims, the Jacob Gould Schurman Professor of Computer Science and Information Science, interim dean of Cornell Bowers CIS, and a Cornell AI Initiative director.

Project Highlights

From enriching the student advising experience to increasing visibility for faculty research, the following solutions inspired audience members to consider generative AI models in addressing ineffective processes in their own departments and teams

Find the faculty members who can talk about [topic].

When asked for faculty members who can answer journalists’ questions about a trending topic, participate with external research collaborators, or respond to a potential donor’s request about funding a niche research project, Dean Meloney and his team in Academic Affairs for Cornell’s SC Johnson College of Business needed a faster and more comprehensive tool than their current solutions. Built in a single semester, the new search tool’s results are faster and produce more comprehensive results that include new faculty members and recent work that may have been overlooked by the previous processes.

Help me prepare for advising this student.

Sharing the student advising load ensures students receive timely appointments while reducing the likelihood of a holistic understanding of the student’s goals and challenges. Repeating this information to a new advisor in each appointment is frustrating for students, but advisors need that information to provide the best advice. Vice Provost for Undergraduate Education Lisa Nishii and her team tackled this challenge by creating an AI-driven dashboard for advisors to browse before meeting their student. The dashboard displays the most relevant details needed to grasp a quick history, with links to dive deeper into additional details as needed.

What do you remember from our last lecture? 

Earth and Atmospheric Sciences professor Toby Ault needs to know what his students recall from their previous lecture or assignment before he can build on that knowledge. In small classes, it is easy to use a Socratic conversation to reveal their grasp of that material at the start of class. Attempting individual Socratic conversations across a large class proved to be an inefficient use of Ault’s and the students’ time. Working with Marty Sullivan—who double-hats as a Ph.D. student and a Cornell IT DevOps engineer--Ault envisioned an AI-driven chatbot that could conduct Socratic conversations then highlight previous concepts needing review before he progressed to new material. With the time saved, Ault wrote more grant proposals.

Does this program meet the student’s objectives? 

Vice Provost for External Education Paul Krause faced a different kind of student assessment challenge: how well does an eCornell program meet a student’s self-declared objectives? Krause’s team had already incorporated AI models in a variety of challenges like developing and updating programs and recruiting students. The new challenge was leveraging AI to turn pre-learning journey assessments, post-learning journey assessments, and summarized reports into trustworthy insights. Krause said the tool’s initial results were promising and, with a few more iterations, should transform how the team does its impact surveys.

More effective clinical rotation feedback. 

The effectiveness of a supervisor’s or instructor’s feedback is likely to suffer from the human hesitancy to give bad news. For veterinary medicine students attempting to progress through clinical rotations, feedback is critical to their advancement and their success as a professional clinician after graduation. Neurologist Jonathan Wood turned to an AI chatbot to increase the effectiveness of faculty members’ feedback without dehumanizing the process.

Identify current construction and design specs. 

Construction and design projects across Cornell Facilities frequently shift to address challenges like supply chain issues and regulatory changes. For employees and contractors alike, finding and confirming the current specifications for a project required hours of research in the digital library of documents, cross-referenced by numerical codes. To streamline this process, improve compliance, and foster a more dynamic research interface, Andrew Magre, Erik Eshelman, and their team built a generative AI tool that could turn natural language questions into accurate responses with links to the actual documents backing up that answer.

Improve travel reimbursement cycle. 

Sherry Guernsey, Executive Director for the Shared Services Center, knew she needed to help the SSC reduce the turnaround time for travel reimbursement requests and to improve the end user experience. With a very manual review process, leveraging AI technology would be a great step forward. The AI innovators alongside Guernsey’s team identified a routine task to look at using a customized AI assistant. Currently, this is in a test environment where they will continue to add more tasks which will enable the SSC travel team to spend more of their time and attention on high-risk transactions.

Complete AI-Assisted web accessibility reviews. 

Annie Heckel, Eric Woods, and Nick Tubbs—part of Cornell’s custom web development team—provide a range of services that include website accessibility reviews for campus partners. Because extensive website reviews blend automated and manual methods to ensure all the university’s mandated web accessibility standards are met, only two team members have the expertise necessary to complete these assessments. Their team’s AI challenge resulted in a proof-of-concept solution enabling anyone in their team to successfully conduct basic accessibility testing, a solution they hope to extend to the rest of the campus in a future iteration.

For additional details about each project, watch the Cornell AI Initiative website or view the recording of the January 22, 2025 presentations.

Photo of Dean Meloney, Academic Affairs, Cornell SC Johnson College of Business, mid-presentation by Luke Stewart - Cornell Bowers CIS