Assignment Refractions
What happens when the disciplinary content stays the same, but the assignment architecture changes?
PRISM holds a disciplinary problem steady while changing Purpose, Role, Challenge, or Meaningful Evidence—and therefore what students are responsible for doing.
One Disciplinary Problem
Imagine a materials science course in which students are learning to make defensible material-selection decisions.
The design problem is the same in every version:
A manufacturer is developing an aluminum bicycle crank for a high-use commuter bicycle. The crank must be light enough for efficient riding, strong enough to withstand repeated loading, resistant to fatigue and corrosion, manufacturable at scale, and affordable enough for a mid-market bicycle.
Students are considering the same set of viable aluminum alloys and the same design criteria in every version.
What changes is the assignment.
Starting Assignment
Select the Best Alloy
You are part of an engineering team selecting an aluminum alloy for a bicycle crank.
Review the candidate alloys and the design requirements provided in class. Compare the materials using the relevant criteria, including strength, fatigue resistance, weight, corrosion resistance, manufacturability, and cost.
Select the alloy you believe is the best choice for the crank.
Submit:
- your selected alloy;
- a brief comparison of the strongest alternatives;
- a short recommendation identifying the criteria that mattered most in your decision.
This is the unrefracted assignment. Students make the recommendation themselves by weighing several viable materials against disciplinary criteria. The disciplinary problem, candidate materials, and design constraints will remain constant. Refraction begins by changing one part of the assignment architecture while holding that common problem steady.
Refraction 1 — Change the Role
Instead of asking students to generate the initial recommendation, give that work to AI.
Evaluate an AI Recommendation
You are part of an engineering team reviewing a proposed aluminum alloy for a bicycle crank.
A generative AI system has reviewed the same candidate alloys and design requirements and recommended one alloy for use in the crank. You will receive its recommendation and supporting explanation.
Evaluate the recommendation against the engineering criteria from the course.
Decide whether you would:
- accept the recommendation;
- modify it; or
- reject it and select another alloy.
Submit:
- the AI recommendation;
- the parts of the recommendation you accept;
- anything you reject or qualify;
- your final alloy recommendation.
The disciplinary problem has not changed. What changes is R — Role. AI now generates the first recommendation, so the student’s intellectual responsibility shifts from producing an answer to evaluating one. The Challenge comes from a plausible recommendation that may be incomplete, poorly weighted, or wrong. What students accept, reject, qualify, and ultimately select becomes the Meaningful Evidence.
Refraction 2 — Change the Role and Introduce Challenge
Students make the decision first. AI enters afterward to challenge it.
Defend a Decision Under Pressure
You are part of an engineering team selecting an aluminum alloy for a bicycle crank.
Review the candidate alloys and design requirements. Select the alloy you would recommend and identify the criteria that most strongly shaped your choice.
Then give your recommendation to the AI tool specified by your instructor. Ask it to act as a skeptical senior engineer reviewing your decision.
The senior engineer should challenge your recommendation by questioning an assumption, identifying a weakness, or arguing for a competing alloy.
Respond to the strongest challenge.
You may:
- maintain your recommendation;
- qualify it;
- revise it; or
- choose a different alloy.
You do not need to change your decision simply because it has been challenged.
Submit:
- your original recommendation;
- the strongest challenge raised by the AI;
- your final recommendation;
- a brief indication of what you maintained, qualified, or changed.
Here R — Role changes again: AI is no longer a recommender but a challenger. At the same time, I — Introduce Challenge becomes more explicit. Students must commit before AI enters, then decide whether their judgment still holds under pressure. The comparison between the original and final recommendation provides Meaningful Evidence of what the student maintains, narrows, qualifies, or revises.
Refraction 3 — Change the Challenge
Students make a recommendation, then one design constraint changes.
Redesign When the Conditions Change
You are part of an engineering team selecting an aluminum alloy for a bicycle crank.
Review the candidate alloys and design requirements. Select an alloy and prepare a brief recommendation.
After you make your recommendation, use the AI tool specified by your instructor to generate one changed condition from this list:
- the cost ceiling is reduced;
- the maximum allowable weight is lowered;
- the fatigue-life requirement increases;
- the manufacturing process changes;
- the crank will now be used in a more corrosive environment.
Do not begin again with a new design.
Determine whether your original alloy still makes sense under the new condition.
Revise only what the changed condition requires.
Submit:
- your original recommendation;
- the changed condition;
- your revised recommendation;
- what changed and what remained the same.
This time the central change is I — Introduce Challenge. The original disciplinary judgment remains in place, but it must now operate under altered conditions. AI’s Role is limited to generating controlled variation. Because students cannot simply restart, the new condition creates consequence: the original recommendation has to be sustained, qualified, or revised. The relationship between the original and revised design becomes the Meaningful Evidence.
Refraction 4 — Change the Purpose
Instead of asking which alloy is best, ask whether enough is known to make a responsible recommendation at all.
Decide What You Still Need to Know
You have been asked to recommend an aluminum alloy for a bicycle crank, but the information available to you is incomplete.
Review the design requirements and candidate materials.
Before selecting an alloy, determine what information is still missing.
Identify the three pieces of additional information that would matter most to a responsible recommendation.
Then use the AI tool specified by your instructor as an inquiry partner. Ask it to help you identify assumptions, missing variables, or questions you may not yet have considered.
Evaluate its suggestions.
Submit:
- the three pieces of missing information you consider most important;
- one additional question raised through your AI inquiry that you believe genuinely matters;
- what, if anything, you are willing to conclude from the information currently available.
Do not select an alloy unless you believe the available evidence supports doing so.
This is a deeper refraction because P — Purpose changes. Students are no longer primarily selecting among materials. They are deciding what evidence is necessary before a responsible engineering judgment can be made. AI supports inquiry rather than recommendation. The incomplete information itself provides the Challenge, and what students identify as missing—or refuse to conclude prematurely—becomes the Meaningful Evidence.
Refraction 5 — Change the Role Again
Instead of recommending a winner, AI expands the possibility space. Students narrow it.
Generate Possibilities, Then Narrow Them
You are part of an engineering team selecting an aluminum alloy for a bicycle crank.
Review the design requirements provided in class.
Use the AI tool specified by your instructor to generate three plausible aluminum-alloy candidates for the crank. Ask it to explain briefly why each might be viable.
Do not ask the AI to select the best option.
Evaluate the three candidates against the engineering criteria from the course.
You may reject all three and introduce another candidate if the generated options are inadequate.
Submit:
- the three AI-generated candidates;
- the candidate or candidates you eliminated;
- your final alloy selection;
- the criteria that determined which options survived your evaluation.
The Purpose returns to material selection, but R — Role changes again. AI generates possibilities without making the consequential judgment. Students remain responsible for narrowing the field. The Challenge comes from discriminating among plausible alternatives, and their selections and rejections provide Meaningful Evidence of the standards that actually governed the decision.
What Changed?
The disciplinary problem stayed constant:
- the same bicycle crank;
- the same candidate materials;
- the same engineering criteria;
- the same disciplinary knowledge.
But the intellectual architecture changed.
Version: Starting assignment
Student responsibility: Select the best alloy
AI role: None
Main challenge: Competing viable materials
Meaningful Evidence: Final selection and criteria
Version: Evaluate an AI Recommendation
Student responsibility: Judge whether a recommendation is defensible
AI role: Recommender
Main challenge: Plausible but potentially flawed advice
Meaningful Evidence: Accept, reject, qualify, modify
Version: Defend a Decision Under Pressure
Student responsibility: Sustain or revise an existing judgment
AI role: Challenger
Main challenge: Critique after commitment
Meaningful Evidence: What remains, changes, or narrows
Version: Redesign When Conditions Change
Student responsibility: Reapply judgment under a new constraint
AI role: Source of variation
Main challenge: Changed design condition
Meaningful Evidence: Original/revised decision
Version: Decide What You Still Need to Know
Student responsibility: Determine whether a decision can responsibly be made
AI role: Inquiry partner
Main challenge: Incomplete information
Meaningful Evidence: Missing evidence and withheld conclusions
Version: Generate Possibilities, Then Narrow Them
Student responsibility: Discriminate among viable options
AI role: Possibility generator
Main challenge: Competing alternatives
Meaningful Evidence: Selection and rejection
That is the generative work of refraction.
The disciplinary object stays steady while one or more PRISM dimensions change, producing an intellectually different assignment.