According to a recent news release, HelioCampus’ AI Insights tool uses a “semantic layer” of baked-in parameters to avoid misunderstandings. For example, the company's Product Manager Craig Rudick said, a user might ask an AI tool for information on tuition revenue.
“Well, did you mean gross tuition revenue, or net tuition revenue? Did you mean tuition and fees, you know, receipts or assessed?” he said. “There’s all these different variants of it.”
Rudick said a faculty or staff member at a university might know in their head that “tuition revenue” usually means “net tuition revenue” and not “gross tuition revenue,” but if the AI doesn’t know, it is programmed to guess, and it might not guess the same thing each time for each user. The chatbot's “semantic layer” allows institutions to input rules to account for imprecise language, making its output reliable even for users who aren’t versed in prompt engineering.
Rudick said the chatbot also simplifies the process of gathering data insights, compared to traditional data dashboards that require a bit more expertise. However, the chatbot does provide the back-end information so that data professionals can verify where the answers came from.
HelioCampus turned to NYIT as the alpha partner for its new AI Insights tool because of a longstanding relationship. Through the partnership, Rudick said, the company hopes to figure out who on campus is most helped by the new chatbot. He said HelioCampus has mostly trained the chatbot to analyze course enrollment data so far, checking for accuracy along the way, and slowly expanding to other areas like retention and student success. The company also added beta partners and will continue to expand to more schools.
“Broadening it out so that the list of that set of questions that can be answered effectively by the AI tool is as broad as possible, that's really where we're focused now,” he said, “But again, it's not worth going broader if you can't keep it accurate.”
