Curriculum Exposure Audit · sample report

Keystone State University

Every course the audit could reach, scored on seven dimensions, with the evidence behind each score shown and the guesses marked as guesses.

Scope: Five colleges, nine departments
Evidence: Public catalog and supplied syllabi
Method version: Rubric 1.0 · scoring prompt 2.0
Run date: September 8, 2026
Current prompt: 2.2 — no change to scoring

How to read this report.

Every score shows its evidence

A score either rests on something the source actually says, or it is marked inferred because the source is silent. The two are never mixed.

Confidence decides what is shown

A course's own result appears only when its evidence reaches moderate confidence or better. Below that it still counts toward department and college figures, but is not shown course by course.

Two independent readings

Each course is scored twice, independently. Where the two passes differ on any score, a third pass decides it by a fixed rule — never by choosing the more convenient answer.

It stops at review

The instrument names courses for review, redesign or investment. It does not recommend eliminating a course or a program, and it is not an instrument for evaluating faculty.

The seven questions asked of every course.

Four ask where AI puts pressure on the course. Three ask where the course builds what AI does not replace. They are scored separately and never netted against each other, because a course can be high on both.

Dimension	Kind	The question it asks of every course
A Task Automation Exposure	risk	Can AI perform, speed up or simulate the tasks students repeatedly do?
B Content Commoditization	risk	Is the course's value mainly in explaining information AI now explains on demand?
C Assessment Vulnerability	risk	Could a student complete the assessed work with undisclosed AI help?
D Labor Market Displacement	risk	How exposed are the occupations this major leads to?
E Human Capability Durability	opportunity	Does the course build judgment, ethics, collaboration or hands-on capability — and assess it?
F AI Augmentation Potential	opportunity	Could well-designed AI use improve learning, feedback or practice here?
G Experiential and Applied Intensity	opportunity	Do students do consequential work under real conditions — labs, field work, placements, clients?

Risk dimensions combine into a displacement risk score (A 30%, B 20%, C 30%, D 20%); opportunity dimensions into a relevance opportunity score (E 30%, F 30%, G 40%). Both run 1–5.

The institution.

146 of 187 catalog courses scored (78%). This is a sample, not a census. Every figure below describes the scored courses only.

3.33

Mean displacement risk

3.46

Mean relevance opportunity

Courses in distinctive strength

Courses in priority redesign

Evidence. 69% of scored courses were read from a syllabus, 31% from the catalog alone. Confidence: 79 high · 44 moderate · 19 low · 4 insufficient.

Reader's note. Risk here is driven primarily by Assessment Vulnerability (mean 3.38) and Task Automation (mean 3.32). Experiential Intensity is the lowest-scoring opportunity dimension (mean 3.06) and is the largest available lever on the opportunity composite. 31% of these courses were scored from catalog evidence only; their Assessment Vulnerability and Experiential Intensity figures are inferred.

Action category	Courses	What it means
Evidence collection	23	The evidence does not yet support a conclusion either way. The next step is to gather it, usually the current syllabus.
Monitor	4	Nothing indicated now. Revisit in the normal review cycle.
Protect and articulate	50	The course builds capabilities AI does not readily replace. Make that value explicit, to students and in program materials.
AI augmentation	3	The course would benefit from deliberate, well-designed AI use in learning or practice.
Assessment redesign	10	Exposure is concentrated in how the course is assessed. Change the assessment, not the subject.
Course redesign	56	Exposure runs through the course's tasks and content, not only its assessment. A faculty-led look at the course design is indicated.

By college.

The institutional averages hide the thing that matters. Education and Health Sciences sit where an institution wants to be. Business does not, and the gap between them is the finding.

College	Scored	Syllabus tier	Risk	Opportunity	Distinctive strength	Priority redesign
Arts & Sciences	21 of 44	10%	3.72	3.42	4	7
Business	42 of 44	90%	4.13	2.94	1	10
Education	15 of 19	80%	1.98	4.13	15	0
Engineering	37 of 44	60%	3.63	3.29	11	4
Health Sciences	31 of 36	87%	2.30	4.04	26	1

By department.

Departments are where a curriculum committee can actually act. Note how far apart Accounting and Nursing sit inside the same institution.

Department	Scored	Syllabus tier	Risk	Opportunity	Distinctive strength	Priority redesign
Accounting	16 of 18	81%	4.33	2.97	0	3
Computer Science	20 of 23	70%	4.08	2.78	0	1
English	12 of 24	0%	3.70	3.42	3	3
History	9 of 20	22%	3.76	3.43	1	4
Kinesiology	10 of 14	70%	3.02	3.89	7	1
Management	26 of 26	96%	4.02	2.93	1	7
Mechanical Engineering	17 of 21	47%	3.13	3.86	11	3
Nursing	21 of 22	95%	1.95	4.12	19	0
Teacher Education	15 of 19	80%	1.98	4.13	15	0

What the evidence supports, course by course.

Four courses shown in full. Each had a syllabus to read as well as a catalog entry — which is what lifted them from a catalog-tier guess to a finding a department can argue with.

ACCT 301 · Intermediate Accounting I

Accounting · high confidence

Course redesign

Risk: 4.4of 5
Opportunity: 2.6of 5
Quadrant: Review
Inferred: none

Evidence. From the catalog alone: 5 of 7 inferred, low confidence, no result of its own. With the syllabus (Fall 2026 — 0 of 7 dimensions inferred, high confidence).

Suggested next move. Most of what students practice week to week — preparing entries, reconciling accounts, building schedules from a trial balance — is work software already does and AI now does faster. Seventy per cent of the grade is take-home problem sets. That combination is why the indicated step is the course design rather than the assessment alone: changing how the work is collected would not change what the work is.

CS 210 · Data Structures and Algorithms

Computer Science · high confidence

Assessment redesign

Risk: 4.1of 5
Opportunity: 3.2of 5
Quadrant: Review
Inferred: none

Evidence. From the catalog alone: 4 of 7 inferred, low confidence, no result of its own. With the syllabus (Fall 2026 — 0 of 7 dimensions inferred, high confidence).

Suggested next move. The subject matter holds its value; the grade may not. Four unproctored programming assignments carry 60 per cent, and each is a task a model completes in seconds. The in-class design critique already in the syllabus is the part that still shows what a student can do unaided, and it carries 10 per cent. Exposure here is concentrated in the assessment scheme, which is the narrower and cheaper thing to change.

NURS 340 · Clinical Practicum II

Nursing · high confidence

Protect and articulate

Risk: 1.9of 5
Opportunity: 4.5of 5
Quadrant: Distinctive strength
Inferred: F

Evidence. From the catalog alone: 4 of 7 inferred, low confidence, no result of its own. With the syllabus (Fall 2026 — 1 of 7 dimensions inferred, high confidence).

Suggested next move. Students are assessed on supervised patient care, observed directly by a preceptor against a competency checklist. There is no route through this course that a model can take. The step indicated is not a change to the course but a claim the college should be making out loud, in program materials and to applicants. Whether AI tools are used in care planning was not stated and is worth recording.

EDUC 415 · Student Teaching Seminar

Teacher Education · moderate confidence

Protect and articulate

Risk: 2.1of 5
Opportunity: 4.2of 5
Quadrant: Distinctive strength
Inferred: C, F

Evidence. From the catalog alone: 4 of 7 inferred, low confidence, no result of its own. With the syllabus (Spring 2026 — 2 of 7 dimensions inferred, moderate confidence).

Suggested next move. The seminar runs alongside a placement, and the assessed work is reflective practice on lessons the student actually taught — grounded in events a model has no access to. Two points could not be read from the syllabus: how the reflective journals are graded, and whether the course says anything about AI use in lesson planning. Both would raise confidence to high.

A catalog tells us what a course covers. A syllabus tells us how it is graded.

That difference decides how much of this report is evidence and how much is estimate. A catalog is written for accreditation and recruitment; it never describes the assignments, the weightings or the exam format — which is exactly where AI exposure concentrates. So wherever a course was read from the catalog alone, Assessment Vulnerability is marked inferred and its confidence cannot reach high.

Across this run, 31% of courses were scored from the catalog alone. Every one of them carries that mark. You can always tell which findings rest on evidence and which on inference, because the report says so on the course.

What the audit will never do.

Recommend eliminating, consolidating or closing a course or a program.
Evaluate a faculty member. Results are not an input to personnel, promotion, tenure, workload or contract decisions, and we ask for that in writing before faculty review begins.
Turn a machine score into an institutional finding. Nothing becomes a finding until faculty have reviewed it and leadership has approved it.
Claim a course-level result the evidence does not support. Below moderate confidence, the course is flagged for evidence collection instead.

A note on the version.

This run was scored under prompt 2.0. The current prompt is 2.2, and the difference is worth stating rather than glossing: 2.1 and 2.2 changed nothing in the anchors, weights, composites, confidence rules or action rules. They added the workflow by which a faculty reviewer supplies evidence against a named dimension, and a rule that no output field may carry an individual’s name. The rubric is unchanged at 1.0, so every figure above reads the same under either version.

On a live institutional run the rule is stricter: every course is re-scored under one prompt version before faculty review begins, because results from different versions are never mixed. That rule protects an institution’s findings. It does not apply to an illustration, and saying so is more useful than quietly restamping the page.

Terms used in this report.

Score (1–5): Each dimension is scored 1 (very low) to 5 (very high) against fixed written descriptions of each level. On risk dimensions (A–D) higher means more exposed; on opportunity dimensions (E–G) higher means a stronger position.
Catalog tier: Scoring from the public catalog alone: course descriptions, prerequisites, formats and program requirements. The lowest evidence tier; confidence can reach moderate, never high.
Syllabus tier: Scoring from the current syllabus: assignments, assessment scheme, stated outcomes, AI policy. Confidence can reach high.
Inferred: The source says nothing direct about this dimension, so the score comes from course level, format or discipline. At catalog tier, assessment (C) is always inferred, because catalogs do not describe assessment.
Confidence: How far the evidence supports a course's scores: insufficient, low, moderate or high. It falls to low when three or more dimensions are inferred.
Course-level result: A course's own scores and category, shown only at moderate or high confidence.
Aggregate only: Low-confidence scores. They count toward department and college figures but are not shown for the course itself.
Risk and opportunity: The two composite scores, each a weighted average of its dimensions on the same 1–5 scale. Shown separately, never netted against each other — a course can be high on both.
Quadrant: Where a course sits with 3.5 as the line on each composite: distinctive strength (lower risk, higher opportunity), priority redesign (both high), review (higher risk, lower opportunity), stable (both lower).

Your catalog

Keystone is invented. Your catalog is already online.

Which means the same run can start on your courses without anyone sending us a file. Name a college or department, and the argument becomes about your curriculum rather than a fictional one.

Ask about your catalog What this is →