AI Literacy / AI readiness
How do you assess whether a college is ready for AI?
By finding out what your people already do, not by scoring the institution against a maturity model.
A useful AI readiness assessment measures four things: what tools your employees and students are already using, what your data and systems can actually support, whether the people who will decide understand the technology, and what your governance and policy currently permit. It takes six to ten weeks.
What does not work is a maturity-model scorecard that rates the institution one through five on a dozen dimensions. Those produce a number and a slide, and no decision. The finding that changes what an institution does is almost always the shadow-usage inventory - the gap between what leadership believes is happening and what is actually happening.
Why scorecards fail
The market is full of AI readiness frameworks that score an institution across dimensions like strategy, infrastructure, talent, and culture, producing a level between one and five. They are appealing because they are legible: the cabinet gets a chart, the board gets a number, and there is a clear direction of travel.
They fail because a score is not a decision. Being told the institution is at level two on culture does not tell a president what to do on Monday. And the scoring is necessarily self-reported, which means it measures institutional self-perception rather than institutional reality - and those two diverge most in exactly the areas that matter.
The assessments that change behavior are diagnostic rather than evaluative. They ask what is actually happening, and they usually surface something the cabinet did not know.
A maturity score tells you where you rank. It does not tell you what to do on Monday.
The four things worth measuring
Six to ten weeks, ending in decisions rather than a rating.
| Dimension | Method | What it typically reveals |
|---|---|---|
| Actual usage | Anonymous survey plus network and expenditure review | Far more use than leadership expects, concentrated in unexpected departments, frequently with student data in consumer tools. This is the finding that moves institutions. |
| Technical capacity | Systems, data quality, and integration review | That the constraint is rarely AI capability. It is student information system data quality, integration debt, and identity management. |
| Decision-maker literacy | Brief assessment across cabinet, senate leadership, and board | Wide variance within the same body, which explains why AI conversations at that institution keep stalling on definitions. |
| Policy and governance | Document review against actual practice | Existing policy on data, procurement, and academic integrity already covers more than people think, and contradicts practice in specific identifiable places. |
The anonymous survey has to be genuinely anonymous and framed without judgment, or it measures nothing. If employees believe answers could be attributed, they will underreport, and an underreported shadow-usage inventory is worse than none because it produces false confidence.
What the assessment should produce
- Shadow usage inventory by division and tool class
- Data exposure findings, with specific instances
- Technical constraint list, ranked by what blocks what
- Literacy baseline by governance body
- Policy gap analysis against current practice
- Ranked recommendations with owners and sequence
- Cabinet and board presentation
The ranked recommendations are the point. Most assessments end with a list of everything that could be improved, which leaves the institution exactly where it started. What a cabinet needs is the first three things in order, with an owner attached to each and an honest statement of what has to be true for the fourth to matter.
In our experience the first recommendation is almost never a tool or a platform. It is either an interim data-handling guideline, because the exposure findings demand it, or a literacy baseline for the group that has to make the subsequent decisions.
The assessment also produces something less tangible and often more useful: a shared factual basis. A cabinet arguing about AI in the abstract makes no progress. The same cabinet looking at its own usage data makes decisions.
What we look for that institutions miss
- Procurement already in placeAI features arriving inside tools the institution already licenses. Most colleges have deployed AI capability through vendor updates without any review.
- The advising and aid questionWhether any system influencing admissions, advising, or aid packaging uses AI, and what human review exists. Highest exposure, least examined.
- Accessibility conformanceAI-generated content and AI-driven interfaces frequently fail accessibility requirements the institution is legally obligated to meet.
- Adjunct and part-time facultyUsually excluded from surveys and professional development, often the heaviest users, and teaching a large share of sections.
- What students actually experienceInstitutions assess their own readiness and skip the group whose education is affected. Student focus groups routinely contradict faculty assumptions.
Who does this work
Former presidents and system executives who have run diagnostic work inside institutions, alongside CBT’s own leadership.
How long does an AI readiness assessment take?
Six to ten weeks for most single institutions, longer for a multi-college district where each college needs its own usage picture. The survey window is the fixed cost - it needs two to three weeks open to get representative response - and the rest is analysis and consultation.
What does an AI readiness assessment measure?
Four things: what AI tools employees and students are actually using, what your data systems and integrations can support, how well the people who have to decide understand the technology, and what your existing policy and governance already permit or prohibit. The usage inventory is usually the most consequential.
Is an AI maturity model useful?
Rarely, in our experience. Maturity scores are self-reported, which means they measure institutional self-perception rather than practice, and a level rating does not tell a cabinet what to do next. A diagnostic that surfaces actual usage and specific exposure produces decisions; a scorecard produces a slide.
Should the assessment include students?
Yes, and most do not. Students are the most affected group and among the most experienced users, and student focus groups routinely contradict faculty assumptions about what is happening in courses. Excluding them produces an assessment that describes the institution rather than the education.
What usually comes first after the assessment?
Almost never a tool or platform purchase. Typically either interim data-handling guidance, because the exposure findings require an immediate response, or a literacy baseline for the governance bodies that have to make the decisions that follow. Both are cheap and both unblock everything after them.
Find out what is already happening.
Most institutions are surprised by their own usage data. The conversation is more productive once everyone is looking at the same facts.
Related