IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Why UChicago Built Its Own Chatbot Instead of Buying One

The University of Chicago’s proprietary chatbot, PhoenixAI, leverages OpenAI models but serves as an open-ended platform for faculty, students and administrators to experiment and find new use cases.

As artificial intelligence becomes more ingrained in higher education, universities face a choice: to purchase commercial AI services, or build their own? According to the University of Chicago’s Chief Technology Officer Kemal Badur, who spoke at an EDUCAUSE webinar this week, schools risk being left behind if they don’t start somewhere.

“Waiting, I don't feel is an option,” Badur said. “This is not going to settle down. There will not be a time where somebody will release the perfect product that you need, and keeping up is really hard.”

For UChicago, keeping up meant taking matters into their own hands. Instead of purchasing individual AI subscriptions — an approach Badur said was not financially viable for their campus community of around 30,000 — the university created its own generative AI chatbot, PhoenixAI, using OpenAI's GPT-4o as a model. After securing funding from the provost’s office in July 2024, the university had just two months to complete the project before the start of the academic quarter.

The chatbot, which officially launched in September 2024, was developed in just six weeks through a collaboration between UChicago’s IT team, Microsoft and the IT consultant Royal Cyber. Badur credited the university’s familiarity with Microsoft Azure's infrastructure and security measures with enabling the rapid rollout, and said flexibility is an advantage of developing a proprietary platform.

“We can bring these things on board the way we want to and at the cost that we want rather than being beholden to a vendor's timeline to do it for us,” he said. “We have a lot of engagement with users and the campus right now, and we're learning what they want.”

Badur added that a university-specific platform also allows customization to the institution’s needs and goals. Unlike AI tools designed for specific tasks, such as advising students or helping human resources, PhoenixAI was built as an open-ended platform for experimentation. Faculty, students and administrators have been encouraged to explore its capabilities, leading to a diverse range of use cases.

For example, a calculus professor at UChicago recently developed an AI-powered tutor to help students with foundational concepts that she doesn’t necessarily have time to review in a 10-week course. The system provides step-by-step guidance while encouraging students to work through problems on their own.

“It's a very innovative thing to do,” Badur said. “It's not necessarily what you would think of an LLM to do, and she did it without talking to any technical people.”

Additionally, researchers have used the tool to summarize articles, develop basic project plans and integrate disparate data sources for analysis, UChicago Senior Business Systems Analyst Dan Venske said during the webinar. Valerie Archambeau, director of communication technology services, said PhoenixAI can also help with human resources tasks like drafting job descriptions and refining performance goals.

Badur said adoption of PhoenixAI is on par with trends at other institutions, sitting at 15 percent. He sees this as a feature rather than a flaw, as it shows that purchasing licenses for 100 percent of the campus would have left many of them unused, and it allows the school to scale and refine the tool responsibly.

In the future, Badur said the university plans to include search and event-generation features, integrate PhoenixAI with external knowledge sources like Google Drive and Microsoft OneDrive, and also offer more privacy features for personal use and Health Insurance Portability and Accountability Act (HIPAA) compliance.

Beyond technical improvements, UChicago is also investing in AI literacy efforts to ensure its campus community can make the most of PhoenixAI, including a learning module in Canvas rolling out this summer.

Badur said having a chatbot program that is accessible to everyone, and able to be augmented as technologies evolve, has established a baseline of AI maturity that will help protect the university from falling behind.

“In the end, I believe something like this is going to be required for every institution, just like email is now,” he said.