Faculty · The RAFT Framework for Ethical AI Use

A statistical claim under review

Distinguish an observed change from a causal claim and explain percentages accurately.

All redesign examples

Before

Use a tool to analyze the workshop results and write a summary of whether the workshop worked.

What the original hides

A generated summary may replace interpretation while sounding certain about causation.

After

Evaluate a simulated workshop report. Submit your calculations and a 250–350 word conclusion with one limitation and one proposal for better evidence.

Student process

1. Research Independently First: calculate both changes and write what each result does and does not establish before consulting a tool.
2. Augment Your Work: AI may ask questions about your reasoning or critique your interpretation. It may not calculate the submitted answers or draft the conclusion.
3. Fact-Check Everything: reproduce calculations independently. Check whether the design supports causal or population-wide claims.
4. Transform Into Your Own Work: write a qualified conclusion, explain percentage points versus relative change, and propose one design improvement with a limitation.

Evidence to submit

Submit the initial calculation sheet, checked calculations, conclusion, and a brief account of any permitted assistance. Students can request the same critique from a peer.