← All domains
Domain 01 · Domains 1–4
Teaching, Learning & Assessment
Instructional practice, course design, assessment, and classroom AI use within individual courses.
Key AI risks
▹Academic-integrity erosion and invalid assessment
▹Unequal student access to AI tools creates equity gaps
▹Unreliable AI-detection tools produce false accusations
▹Over-surveillance of students through learning analytics
Governed through the RMF Playbook
Each function pairs the outcome to reach for this domain (the Playbook's About) with the Playbook's suggested actions.
Govern
Set the culture and authority for AI in the classroom — what counts as acceptable use, who decides it, and where accountability for course-level AI rests.
Playbook · suggested actions
Publish an acceptable-use policy and course-tier framework; name who owns teaching-AI decisions.
Map
Understand where AI enters each course and what it touches — the tasks, the student data, and the equity context that frame the risk.
Playbook · suggested actions
Require a syllabus AI-use statement and record each tool’s risk tier and the data it touches.
Measure
Judge whether AI use preserves learning and the validity of assessment, and whether it works fairly and accessibly for every student.
Playbook · suggested actions
Pilot with outcome comparisons and run accessibility and bias checks before scaling.
Manage
Act on what you learn — guide faculty, adapt course and assessment design, resolve integrity questions, and revisit expectations as the tools change.
Playbook · suggested actions
Give faculty guidance and a non-AI path, and review course expectations each term.
Worked case studies · the RMF Playbook applied
10 scenarios
Each step is a Playbook suggested action for that function; the evidence line is the transparency & documentation you keep.
AI tutor in a 600-student gateway course
Intro Statistics · Center for Teaching partnership
Faculty want an AI practice-tutor for problem sets but worry about academic integrity and unequal access to paid tools.
Govern
Adopt the AI academic-integrity policy and set the course to "Yellow" (AI permitted with disclosure) in the syllabus statement.
Map
Classify the tutor as assistive / low-risk and document who is affected — a site license means no student pays out of pocket.
Measure
Pilot against a control section, compare learning outcomes, and check the tool for accessibility / UDL conformance.
Manage
Publish disclosure expectations, offer a non-AI practice path, and review at term end.
Outcome — The tutor is approved at the Yellow tier and integrity referrals fall because expectations are explicit.
Evidence: Syllabus AI-use statement + pilot outcome memo
AI-generated practice question bank in nursing
School of Nursing · pathophysiology
An instructor wants AI to draft a large bank of practice quiz questions, raising clinical-accuracy and integrity concerns.
Govern
The acceptable-use policy permits AI drafting only with faculty sign-off before any item reaches students.
Map
Classify as assistive and flag clinical-accuracy risk; the items inform practice, not grades.
Measure
A subject-matter expert reviews every item and tracks the error rate across the drafted bank.
Manage
Only expert-approved items publish; provenance is logged and the bank is revised annually.
Outcome — A vetted practice bank ships with a documented review; two hallucinated items are caught before release.
Evidence: Item-review log + acceptable-use sign-off
Automated draft feedback in first-year writing
Writing Program · composition
The program pilots an AI tool that gives students draft-level feedback before instructor grading.
Govern
Set the tool to Yellow with required disclosure — it advises, the instructor grades.
Map
Classify as operational; document that feedback touches student writing (FERPA) and note access equity.
Measure
Compare revision quality with and without the tool and check feedback for bias against multilingual writers.
Manage
Keep a non-AI feedback path, let instructors override, and review each term.
Outcome — Revision quality rises and a bias check on multilingual writers prompts a prompt change before scale-up.
Evidence: Pilot comparison memo + bias-check notes
Fair handling of suspected AI-written work
Academic Integrity Office
An instructor wants to file an integrity case based mainly on an AI-detector score.
Govern
Policy bars a detector score as sole evidence and guarantees the student a right to respond and appeal.
Map
Classify AI detectors as unreliable and identify elevated false-positive risk for some student groups.
Measure
Require corroborating evidence and audit the detector’s false-positive rate.
Manage
Use a conversation-first process, document it, and route disputes through appeal.
Outcome — No case rests on a detector alone and a likely false accusation is avoided.
Evidence: Integrity case file + detector-limitations guidance
Rolling Green/Yellow/Red across a department
History Department · curriculum committee
A department wants consistent course-level AI expectations instead of ad-hoc, instructor-by-instructor rules.
Govern
Adopt the institutional Green/Yellow/Red framework as departmental policy.
Map
Each course classifies its AI stance and states it explicitly on the syllabus.
Measure
Audit syllabi for a compliant statement and survey students on whether expectations are clear.
Manage
The committee updates guidance yearly and adjudicates edge cases.
Outcome — Every syllabus carries a clear AI-use statement and student confusion drops sharply.
Evidence: Syllabus audit + course AI-use statements
AI accommodations for students with disabilities
Disability Services · required course tool
A tool the course requires must work for students using screen readers and extended time.
Govern
The accessibility / UDL standard requires conformance for any mandated AI tool.
Map
Identify affected students and classify the tool as required, which raises the bar.
Measure
Run an accessibility conformance test and gather feedback from affected students.
Manage
Remediate the tool or provide an equivalent path before it becomes required.
Outcome — The tool meets accessibility requirements before rollout, with an equivalent path on record.
Evidence: Accessibility conformance report
Grading code at scale with AI assistance
Computer Science · intro programming
TAs want AI help grading thousands of coding submissions quickly and consistently.
Govern
The AI-assisted grading policy requires human review of every AI-suggested grade.
Map
Classify as operational and document that grades are consequential and the data is student work.
Measure
Spot-check AI–human grade agreement and test for disparities across sections.
Manage
TAs confirm every grade, students can appeal, and discrepancies are logged.
Outcome — Grading speeds up while humans own every final grade and agreement is tracked.
Evidence: Grade-agreement audit + appeals log
Equity when students use AI unevenly in teams
Business · team capstone
Some students on a team have paid AI tools and others don’t, skewing contributions and grades.
Govern
Equitable-assessment guidance explicitly addresses differential access to AI tools.
Map
Identify access gaps within teams and classify it as an equity risk.
Measure
Provide an institutional tool so access is even and assess individual contribution.
Manage
Site-license the tool and design the assessment to credit individual work.
Outcome — All students get the same tool and grading reflects individual contribution rather than tool budget.
Evidence: Tool-access record + individual-contribution rubric
Designing AI-resistant authentic assessments
Center for Teaching · faculty cohort
Faculty want assessments that stay valid even when students have full access to AI.
Govern
Apply the authentic-assessment design standard for AI-prevalent environments.
Map
Identify which outcomes AI can shortcut and scope the redesign accordingly.
Measure
Pilot the redesigned assessments and check both validity and grading workload.
Manage
Fund redesign grants, share exemplars, and review the approach annually.
Outcome — Assessments emphasize process and application, and validity holds with AI present.
Evidence: Redesigned assessment set + validity notes
AI conversation practice in a language course
Modern Languages · intermediate Spanish
An instructor adds an AI chat partner for low-stakes speaking and writing practice.
Govern
Set the tool to Yellow with disclosure and no assessment stakes.
Map
Classify as assistive and document the student voice and text data it collects.
Measure
Check the tool’s language accuracy and privacy terms and gather learner feedback.
Manage
Offer a human conversation-partner alternative and review the vendor terms.
Outcome — Students get abundant low-stakes practice and a privacy review clears the tool for use.
Evidence: Privacy review + learner feedback
Risk-tier examples
Assistive
AI writing assistants used for drafting and editing.
Operational
AI-assisted grading tools that give feedback on student work.
Consequential
AI informing decisions about academic standing.
Tools & artifacts
AI system inventory entry
Green/Yellow/Red syllabus template
Accessibility / UDL checklist
Pilot evaluation rubric
Key controls & instruments
AI academic-integrity policy
Green/Yellow/Red course-use tiers
Authentic assessment design standards
Syllabus AI-use statement requirement