AI Deep Dive: AI-Infused Curriculum and Program Design

Primary Audience: IR directors and analysts, strategic planning and institutional effectiveness leaders, select cabinet members 

This session reports from practice. The IR office at a small private university has used AI to change what the function produces, not to do the old job faster. Professional associations in the field describe a required shift from compliance reporting and retrospective benchmarking toward strategic decision support. This session shows that shift as built, working artifacts in current use. 

The first terrain uses AI to turn reporting and operations into software. Raw data becomes a polished, multi-page, customized report on demand, in minutes rather than days. A compliance tracker maintains itself from documents the office already receives, with no manual entry of items or dates, and progress reports and planning dashboards generate from the same tracker. Static reports gain interactive companions that open in any browser, where users explore the data instead of reading frozen tables and where complex dynamics, such as students moving between majors, become visible at a glance. 

The second terrain uses AI to turn free public data into a shared decision-support service. Questions that institutions pay vendors to answer are answered in house, on a schedule: grant opportunities screened for fit, recruitment markets mapped school by school, labor-market demand profiled program by program. Each product arrives as the short list that matters, not the raw feed, and judgment stays with the consuming department. 

The third terrain is the least charted: AI that treats institutional text as measurable data. Student comments, learning outcomes, and strategic plans hold signal that offices shelve as “qualitative.” The session shows that text turned into defensible numbers, tracking how meaning shifts across years of documents and scoring how well one set of texts aligns with another. The measures are auditable and reproducible, and no mathematics background is required to use them or to explain them to a cabinet. 

All of this runs without exposing institutional data to consumer AI chatbots or third-party cloud AI services. Data stays on institutional devices, models run locally, and results come from scripted pipelines rather than from records pasted into a chat window. Nothing here is a black box: the methodology is transparent, auditable, and reproducible, and every figure traces to code that can be rerun on demand. 

The session is demonstration driven, and each demonstration includes the method behind it. Participants leave with a map of where AI extends IR capacity and use cases worth reproducing at other institutions. Bring a question your institution pays a vendor to answer, a corpus of institutional text shelved as unmeasurable, or a recent board or effectiveness document, and we will trace which workflow would take it on.