About this framework
Campuses approve AI and pay for it, but few can say which uses are worth it. This framework gives everyone one consistent way to find out and act.
Why it exists
This site grew out of my own observation: there are many claims about the value of AI in higher education, but I rarely come across empirical evidence behind them. Once AI is approved and running, the simple question goes unanswered: is it worth it, and what should we do next?
- Use is mistaken for value. Logins and licenses show access, not improvement.
- Capacity is mistaken for cash savings. Freed hours cut costs only when spending falls.
- Renewals run on habit. With no baseline, no one can show a change.
- Each unit judges differently. Results can’t be compared across campus.
- Risk is checked once. Quality, equity, and vendor changes drift after approval.
What it is
An evaluation and decision framework for any approved AI use, from one person’s task to a campus-wide platform.
Measure Cost, Use, and Outcomes against a before-AI baseline. Run five checks that can stop a use. Then make one of five decisions: continue, improve, expand, restrict, or retire, with an owner and a date.
Effort scales with the stakes, from a quick self-check to a formal review. Decisions feed the budget and program reviews the campus already holds.
It pairs with the AI Strategic Compass, which decides what to start, and the Campus AI Registry, which records each use. How they fit together
Who it’s for
Anyone using AI at work, team and unit leads running pilots, IT and AI governance leads, and budget offices. Start with one use; no campus-wide system needed.
Joe Sabado
- Share and adapt it for your campus.
- Credit the author.
- Non-commercial only. Ask before using it in paid products or services.
- Share adaptations under the same license.
Credit line: “AI Cost, Use, and Outcomes Framework by Joe Sabado, licensed under CC BY-NC-SA 4.0.”
Status and evidence base
- Working concept, version 1.0, last updated October 2026.
- Evidence base: see References; established evaluation practice (baselines, comparators, attribution) and the distinction budget offices draw between cash and capacity savings, applied to AI.
- Case studies are illustrative. Figures are invented to show the method. None are field-tested or independently evaluated yet.
- Stress-tested, not validated: the 500-case inventory checks that the rules behave sensibly; it is not campus data.
- Boundaries: it complements AI governance, procurement, privacy, and accessibility review. It doesn’t replace them or decide what is legal.
This is a working concept for campuses to consider and adapt. If you use it, share how you’re using it and how to improve it.
Accessibility
Every page works by keyboard and on small screens. The contents drawer closes with Esc. If anything gets in your way, email joepsabado@gmail.com.