Institutions · The RAFT Framework for Ethical AI Use

RAFT^™ leadership brief

Purpose scope and evidence

Open resource · No email required

The RAFT Framework for Ethical AI Use, by Dr. Chase Cookson, gives students and faculty a common process for learning-centered AI use. Consider a bounded teaching pilot before any institution-wide adoption claim.

Purpose

RAFT translates a broad expectation—use AI responsibly—into actions learners can explain while completing an assignment. Its central question is: Am I using AI to enhance my learning, or to bypass it?

• Research Independently First: gather credible sources and form an initial analysis.
• Augment Your Work: seek questions, gaps, and counterarguments rather than a finished submission.
• Fact-Check Everything: verify retained claims, citations, calculations, and summaries.
• Transform Into Your Own Work: build the final product through your own reasoning and be able to defend it.

Scope

Use RAFT in assignment instructions, classroom activities, feedback conversations, and process documentation. Begin with the learning goal, applicable rules, and the learner’s independent starting point. Transparency, authenticity, and responsibility surround all four steps.

What leaders would be approving

A limited instructional pilot with named owners, an accessible no-AI route, local-policy review, faculty preparation, and an evaluation plan. Adoption does not require purchasing an AI product or collecting student conversations.

What evidence supports the decision

The current RAFT materials establish a coherent design and a set of usable teaching resources. They do not supply a RAFT-specific impact evaluation. This brief makes no claim that RAFT has been proven to improve learning, reduce misconduct, or produce institution-wide outcomes.

A practical evidence plan

Before launch, choose a small number of learning criteria and a comparable independent task. Sample source checking, reasoning, and ability to explain a decision without AI. Record the task conditions and rubric. Add student and faculty feedback about ambiguity, access, and workload.

Report counts and limitations alongside results. Confidence or satisfaction is useful implementation feedback but is not evidence of learning on its own. Consult the institution’s evaluation or research office before treating routine teaching data as publishable research.

Sources and claim boundaries

This brief derives from the RAFT Content Standard and the local Collateral and Services Roadmap. These are design and planning sources, not independent validation. External alignment, when discussed, indicates complementary principles and does not imply endorsement or equivalence.

A bounded pilot decision

Start with a small set of willing instructors and assignments where students must explain evidence and decisions. Choose a timeframe that allows a baseline task, an introduced RAFT task, and an independent follow-up.

Continue revise or stop

Continue only when the pilot can be delivered within the agreed workload and students can show the targeted reasoning. Revise when instructions or evidence requirements create confusion. Pause when access barriers, privacy problems, or unacceptable burden remain unresolved.

Questions for leadership

• Which learning problem are we trying to address?
• What would count as a useful improvement, and how will we check it independently?
• Who can authorize local adaptations and resolve conflicting rules?
• What evidence would cause us to stop or change course?

Use the Campus Readiness and Pilot Planning Workbook to record these decisions. Use the Policy to Practice Mapping Guide to connect actual policy clauses to assignment-level actions.