Cognitive extension starts with a better challenge.

01Curriculum connection

Characterization, motivation, close reading, and textual evidence.

02Challenge

Can your interpretation survive a character who actively resists it?

03Cognitive extension

Sustain and revise competing explanations of a character. AI makes a responsive countervoice available; students judge whether its persuasive self-defense is supported by the text.

04Preparation and AI role

Select a significant decision and two or three relevant passages. AI role-plays the character, then becomes a skeptical reader. Keep the text available as the authority for every claim.

05Learning sequence

1. Write an initial claim explaining the decision and support it with specific passages.
2. Give AI the passages and interview the character. Probe alternative motives, evasions, and contradictions.
3. Pause after a convincing response. Mark what is supported by the text, inferred, or invented.
4. Switch roles. Defend the character’s explanation while AI challenges it as a skeptical reader.
5. Revise your interpretation to address the strongest objection and identify what remains unknowable.

07Evidence of learning

A revised interpretation with textual support, one serious counterargument, and an explanation of why a claim was retained or changed.

01Curriculum connection

Literary conflict, theme, values, and reasoned interpretation.

02Challenge

How can the same facts support different judgments about a character’s decision?

03Cognitive extension

Reason within and across different value systems. AI sustains three perspectives under questioning; students identify the priorities and sacrifices underlying each judgment.

04Preparation and AI role

Choose a literary dilemma and establish its facts. AI represents loyalty, individual freedom, and collective welfare as distinct priorities, without presenting them as complete moral theories.

05Learning sequence

1. Record the facts and make an initial judgment about the character’s decision.
2. Question the three AI perspectives separately about those same facts.
3. Identify each perspective’s strongest argument and the cost it is willing to accept.
4. Introduce a complication from the text. Track whether each perspective remains consistent or changes its reasoning.
5. Formulate your own judgment, naming your value priority and the objection you find hardest to answer.

07Evidence of learning

A judgment supported by textual evidence, with an accurate account of an opposing position and its strongest objection.

01Curriculum connection

Point of view, narrative attention, characterization, and creative interpretation.

02Challenge

What changes when a character the story barely notices becomes its center?

03Cognitive extension

Reconstruct possible perspectives while monitoring uncertainty. AI offers contrasting, text-compatible possibilities; students examine how their own narrative choices redistribute sympathy and responsibility.

04Preparation and AI role

Choose a secondary character and a scene. Separate established facts from unknowns. AI produces imaginative reconstructions, then questions the student’s version.

05Learning sequence

1. List what the text establishes about the character and what it leaves unknown.
2. Ask AI for two contrasting possible experiences of the scene that preserve those facts.
3. Annotate both accounts as textual evidence, reasonable inference, or invention.
4. Write your own version of the scene with the secondary character at its center.
5. Discuss with AI how the rewrite changes judgment of the protagonist. Explain what it reveals and what it may impose on the original.

07Evidence of learning

A rewritten scene with annotations and a commentary that distinguishes supported insight from speculative reconstruction.

01Curriculum connection

Close reading, competing interpretations, argument, and rhetorical language.

02Challenge

Can you distinguish an eloquent interpretation from one that withstands scrutiny?

03Cognitive extension

Compare explanations against resistant evidence rather than fluency. AI creates and defends rival readings; students construct criteria for judging them and strengthen an initially unwelcome alternative.

04Preparation and AI role

Select a short ambiguous passage. AI generates rival readings, including a deliberate overstatement. The teacher and students verify evidence against the passage; AI’s labels are not an answer key.

05Learning sequence

1. Interpret the passage independently and record your reasons.
2. Request three readings: one similar to yours, a strong alternative, and an eloquent overstatement, initially unlabeled.
3. Interrogate each reading. Demand textual support and find details it struggles to explain.
4. Strengthen the interpretation you initially found least convincing before ranking the readings.
5. Ask AI to reveal its construction choices. Evaluate that explanation against your own evidence-based judgment.

07Evidence of learning

A ranked evaluation showing textual coverage, weaknesses, and the strongest challenge to the preferred reading.

01Curriculum connection

Motion, gravity, relationships between variables, and experimental design.

02Challenge

How can you discover a hidden physical relationship through experiments?

03Cognitive extension

Infer a rule and design tests that distinguish models. AI serves as a responsive research partner, helping students confront the predictions and limitations of their own hypotheses.

04Preparation and AI role

Use a teacher-checked simulation with a fixed hidden rule, such as an unfamiliar gravitational acceleration. Establish controls, units, and accessible measurements. All numerical results must come from the simulation; AI must not improvise data.

05Learning sequence

1. Observe a few simulated events and propose two possible explanations.
2. Work with AI to turn each explanation into a testable prediction without revealing the hidden rule.
3. Choose an experiment for which the explanations predict different results. Record predictions before running it.
4. Use the result to revise or reject a model and choose the next experiment.
5. Predict an untested condition, run it, and explain the scope and limits of your proposed relationship.

07Evidence of learning

An investigation log with competing hypotheses, a discriminating test, a revised relationship, and a prediction checked on new data.

01Curriculum connection

Energy transfers, work, power, efficiency, and conservation of energy.

02Challenge

Can you turn an impossible energy claim into a useful, physically feasible design?

03Cognitive extension

Track evolving claims and changing system boundaries. AI provides a persistent inventor whose revisions force students to reapply conservation principles instead of identifying a flaw only once.

04Preparation and AI role

Use a fictional wheel-driven generator claimed to power its bicycle indefinitely. AI plays the inventor. Students use teacher-provided or independently checked values for any numerical feasibility estimate.

05Learning sequence

1. Interrogate the pitch and draw a system diagram showing all energy inputs, transfers, and outputs.
2. Request quantitative claims and use calculations to identify a contradiction or missing energy source.
3. Ask the inventor to revise the proposal. Track every changed assumption.
4. Determine whether the revision resolves the problem or moves it outside the stated system boundary.
5. Redesign the device for a realistic purpose, such as powering a small light, and estimate its limits.

07Evidence of learning

An annotated chain of claims and revisions, an energy diagram, and a feasible redesign supported by a calculation with units.

01Curriculum connection

Position–time graphs, speed, acceleration, measurement, and uncertainty.

02Challenge

What evidence would distinguish two models that both fit the observations?

03Cognitive extension

Allocate a limited measurement budget to reduce uncertainty. AI helps students sustain the strongest version of each model while students decide which observations would be decisive.

04Preparation and AI role

Prepare a short toy-car dataset with stated position uncertainty, compatible with constant speed and small acceleration. Use a repeatable setup or teacher-checked simulator for additional measurements.

05Learning sequence

1. Plot the observations and examine both constant speed and small acceleration as explanations.
2. Ask AI to help articulate the strongest case for each without choosing a winner.
3. Choose only three additional measurements, specifying their timing and required precision.
4. Record each model’s predictions before collecting the new data.
5. Compare the results with the prediction ranges and decide what can be concluded, including whether the models remain indistinguishable.

07Evidence of learning

A graph, a justified measurement plan, predictions for both models, and a conclusion proportionate to the evidence.

01Curriculum connection

Gravity, projectile motion, forces, and modeling assumptions.

02Challenge

Which parts of a delivery-system design survive a change of environment?

03Cognitive extension

Transfer principles across unfamiliar conditions and reason backward from constraints. AI acts as mission control, introducing changes that expose which assumptions the original design depends on.

04Preparation and AI role

Specify a gap, package mass, launch and landing heights, and acceptable landing speed. Use a paper or simulated design. Begin with a manageable model, such as negligible air resistance, and keep numerical conditions explicit.

05Learning sequence

1. Design a package-delivery system for the initial conditions and predict its trajectory and landing speed.
2. Ask AI to introduce one defined environmental change, such as lower gravity.
3. Identify what still applies and predict what must change before recalculating.
4. Revise the design, then address one additional constraint, such as a maximum launch speed.
5. Compare the designs and identify the principles that transfer across environments and the assumptions that do not.

07Evidence of learning

Two or three designs with calculations and an explanation separating general principles from environment-specific assumptions.

01Curriculum connection

Variables, conditionals, lists, testing, and allocation rules.

02Challenge

What happens when people change their behavior in response to your algorithm?

03Cognitive extension

Reason about feedback between rules and strategic behavior. AI generates contrasting fictional users across successive revisions; students evaluate incentives and trade-offs through executable code.

04Preparation and AI role

Choose a limited resource, such as club places, and define fictional applicant fields. Keep the program within familiar Python concepts. AI scenarios illustrate possibilities, not predictions about real populations.

05Learning sequence

1. State the allocation goal and write an explicit rule before coding it.
2. Ask AI for contrasting applicants, including people who adapt strategically to the rule.
3. Predict the outcomes, run the program, and identify unintended effects.
4. Revise the rule and test a second round of AI-generated adaptation.
5. Defend a final version while identifying a trade-off and an exploitation strategy that remain unresolved.

07Evidence of learning

A working program, tests from two rounds, and an explanation of how the rule changes incentives and distributes opportunities.

01Curriculum connection

Boolean logic, string processing, specifications, and ambiguous input.

02Challenge

Can an intuitive idea of a duplicate be made precise enough for a program?

03Cognitive extension

Explore the gap between meaning and formal rules. AI progressively generates boundary cases; students decide which ambiguities code can resolve and which require more information or human review.

04Preparation and AI role

Use fictional registration records and familiar string operations. AI acts as a literal-minded tester. Distinguish duplicate records from different people who share a name.

05Learning sequence

1. Define “duplicate registration” and implement an initial check.
2. Ask AI for progressively harder cases involving capitalization, spaces, nicknames, shared names, and misspellings.
3. Predict and test each outcome, deciding whether the available information is sufficient.
4. Revise the specification and program; introduce a “needs review” outcome where appropriate.
5. Identify what extra information would resolve an ambiguous case and explain the limits of automatic matching.

07Evidence of learning

A duplicate-checking program, revealing test cases, an evolving specification, and a boundary statement for uncertain decisions.

01Curriculum connection

Loops, lists, randomness, simulations, and controlled comparisons.

02Challenge

Can different rules produce the same visible group pattern?

03Cognitive extension

Distinguish outcomes from causal mechanisms. AI proposes a rival explanation; students build and intervene in two simulations to find evidence that separates them.

04Preparation and AI role

Use fictional agents choosing between two options. One model uses imitation; another uses independent preferences. Set clear update rules and compare multiple runs, recording random seeds where used.

05Learning sequence

1. Build the imitation model and record when one choice becomes dominant.
2. Explain the pattern, then ask AI for a different mechanism that could produce a similar outcome.
3. Implement an independent-preference model using familiar Python constructs.
4. Run repeated trials to find conditions in which both models produce similar patterns.
5. Change an input or condition, such as initial choices, to separate their predictions. Explain what the comparison establishes about the models only.

07Evidence of learning

Two simulations, results from repeated runs, and an intervention showing why a shared pattern does not establish a shared cause.

01Curriculum connection

Functions, conditionals, loops, specification, testing, and tracing.

02Challenge

What is the smallest input that disproves a program’s claim to work?

03Cognitive extension

Use counterexamples to expose a general claim’s hidden assumption. AI defends and discusses code; students verify behavior through execution and reduce failures to their essential structure.

04Preparation and AI role

Provide a short AI-generated function, checked by the teacher to contain a discoverable flaw, such as finding the second-largest distinct number. Specify the expected behavior for fewer than two distinct values.

05Learning sequence

1. Predict the function’s behavior on ordinary examples before running it.
2. Interrogate AI’s defense of the logic and construct a potentially failing input.
3. Run or trace the code to establish the failure, then remove unnecessary input elements while preserving it.
4. Explain the broken assumption and repair the function independently.
5. Test an AI-proposed challenge case and rerun earlier tests. Distinguish remaining bugs from undefined requirements.

07Evidence of learning

A minimal counterexample, a trace, corrected code, and tests explaining the improvement and its verification limits.

01Curriculum connection

Historical context, primary sources, contingency, and the distinction between contemporary knowledge and hindsight.

02Challenge

Can you defend a historical decision without knowing how events will unfold?

03Cognitive extension

Reason from incomplete evidence across several plausible futures. AI sustains a historically bounded advisory exchange; students assess what could reasonably have been known and distinguish a sound decision process from a favorable outcome.

04Preparation and AI role

Choose a decision from the current unit, such as a reform organization choosing a strategy. Prepare a dated source packet and a later-evidence packet. Use an AI conversation with only the initial packet; instruct it not to introduce later knowledge. Students audit any leakage.

05Learning sequence

1. Read the initial sources. Record available choices, constraints, uncertainties, and a provisional recommendation.
2. Ask AI to represent two contrasting advisory positions supported by the packet. Interrogate their assumptions and requests for evidence.
3. Develop two plausible future scenarios. Identify which uncertainties could reverse your recommendation; label all scenarios as hypothetical.
4. Commit to a decision and its reasons before opening the later-evidence packet.
5. Compare the decision with subsequent evidence. Explain whether you would revise the reasoning, the decision, or only your expectations.

07Evidence of learning

A dated decision memo, two conditional scenarios, and a retrospective explaining what the later evidence changes and what it does not.

01Curriculum connection

Sourcing, corroboration, archival silences, perspective, and historical argument.

02Challenge

How much of a historical explanation depends on which records survive and which you encounter first?

03Cognitive extension

Investigate the relationship between evidence selection and historical interpretation. AI builds competing source-bounded arguments; students identify archival gaps without treating imagined voices as recovered evidence.

04Preparation and AI role

Prepare two small sets of authentic sources on one unit topic, such as a labor dispute or reform campaign. Give the sets different perspectives and identify authors, dates, audiences, and provenance. Reserve the second set for later disclosure.

05Learning sequence

1. Use the first packet to write a short account and assign a confidence level to its central claim.
2. Ask AI for two rival accounts from the same packet. Trace each claim to a source or mark it as inference.
3. Open the second packet. Test which earlier claims survive, require qualification, or collapse.
4. Have AI interrogate your revised account for whose actions, interests, or experiences remain poorly documented.
5. Write a synthesis and a targeted archival research request specifying the kind of missing evidence that would matter most.

07Evidence of learning

An initial account, a revised synthesis with source references, and a precise request for missing evidence that could change the interpretation.

01Curriculum connection

Causation, chronology, agency, institutions, and the interpretation of historical change.

02Challenge

How can you distinguish a compelling cause from a factor that merely accompanied an outcome?

03Cognitive extension

Compare interacting causal explanations and test their implications. AI sustains rival models while students investigate mechanisms, sequence, and counterevidence rather than settling for a single persuasive narrative.

04Preparation and AI role

Select an outcome from the current unit, such as passage of a reform. Supply a verified chronology and sources representing several relevant actors. AI may help formulate counterfactual tests, but hypothetical outcomes are not historical evidence.

05Learning sequence

1. Build an initial causal map linking actors, conditions, decisions, and the selected outcome. Explain each proposed connection.
2. Ask AI to build the strongest rival model from the same sources, emphasizing a different causal mechanism.
3. Identify an observable implication of each model: what evidence should exist if that mechanism mattered?
4. Search the supplied sources for support and contradiction. Consider a minimal counterfactual change to expose an assumption, without claiming to prove an alternative history.
5. Revise the map to show interactions and distinguish enabling conditions, triggers, and limits of the explanation.

07Evidence of learning

Two causal maps and a defended synthesis that explains a mechanism, addresses counterevidence, and identifies an unresolved causal question.

01Curriculum connection

Historical statistics, change over time, distribution, and the relationship between evidence and evaluative claims.

02Challenge

Can an era look like progress in one defensible account and deterioration in another?

03Cognitive extension

Integrate quantitative and qualitative evidence across different definitions of improvement. AI stages a methodological dispute; students examine whose experience a metric represents and how an evaluation depends on its scale and timeframe.

04Preparation and AI role

Prepare a sourced packet for a period in the current unit, such as industrialization. Include a small, comparable dataset and contrasting firsthand documents. Clarify units, population coverage, and whether monetary values are comparable over time.

05Learning sequence

1. Define “progress” and select two indicators before examining the full packet.
2. Calculate or graph the indicators and compare their trends with the firsthand evidence.
3. Ask AI to defend two opposing evaluations using only the packet. Require it to identify the populations and timeframes behind its claims.
4. Change one analytical choice, such as subgroup, start date, or indicator. Test whether the conclusion changes and why.
5. Write a qualified judgment that identifies gains, costs, distribution, and the evidence needed to resolve remaining uncertainty.

07Evidence of learning

An annotated graph or data table and a historical judgment explaining how results depend on measurement choices and whose experience is included.

01Curriculum connection

Particle models, solutions, electrical conductivity, and the link between macroscopic observations and microscopic explanations.

02Challenge

How can you infer what particles are doing when you cannot observe them directly?

03Cognitive extension

Move among observations, particle representations, and testable models. AI helps sustain competing microscopic explanations; students design a test that can distinguish them rather than accepting a plausible particle story.

04Preparation and AI role

Prepare teacher-checked results for several coded aqueous samples with controlled concentrations and conductivity measurements. Include solubility observations where useful. Use supplied results or teacher-approved measurements; chemical identities and instrument limitations remain explicit in the teacher key.

05Learning sequence

1. Examine the initial observations and draw two possible particle models for a coded sample.
2. Ask AI to challenge how each model accounts for dissolving and conductivity. Separate observation from interpretation.
3. Choose an additional available test or comparison and predict the outcome under each model.
4. Reveal or collect the result. Revise the particle diagram and explain which model is weakened or remains viable.
5. Predict a new sample or condition, then check it against the prepared evidence. State what the data cannot establish about identity.

07Evidence of learning

Initial and revised particle diagrams, a discriminating test, and an explanation connecting visible evidence to a microscopic model.

01Curriculum connection

Bonding, molecular structure, intermolecular interactions, and structure–property relationships.

02Challenge

Can you choose a material when improving one property may undermine another?

03Cognitive extension

Reason across molecular structure, observable properties, and design constraints. AI introduces consequential changes to a design brief; students distinguish chemically supported predictions from attractive but untested claims.

04Preparation and AI role

Use a teacher-curated set of candidate materials for a fictional protective package. Supply structural representations and measured property data, including relevant limitations. Restrict the analysis to bonding concepts already studied; this is a design investigation, not a synthesis task.

05Learning sequence

1. Choose a material for an initial brief, such as stiffness, water resistance, and low mass. State priorities and justify the choice.
2. Explain two properties through structure or interactions, marking where that explanation is incomplete.
3. Ask AI to introduce one plausible new constraint, such as a different service temperature or a requirement to recover the material.
4. Reassess the candidates using the same data. Predict a property trade-off and propose a test that could reveal it.
5. Defend a revised choice or combination, explicitly identifying the most important unknown rather than assuming a perfect material exists.

07Evidence of learning

A candidate comparison, an annotated structure–property explanation, and a revised design with a justified trade-off and proposed verification test.

01Curriculum connection

Reaction rates, collision models, experimental control, and interpreting measurements over time.

02Challenge

What evidence would reveal why two apparently similar reactions proceed at different rates?

03Cognitive extension

Design informative experiments under uncertainty. AI develops rival explanations from disclosed conditions; students decide which variable to isolate and whether the measured signal actually represents reaction progress.

04Preparation and AI role

Prepare teacher-checked rate curves from one familiar reaction or a validated simulator. Include explicit conditions, measurement uncertainty, and one initially undisclosed difference. Provide a defined menu of additional trials. Use numerical data from the experiment or simulator, not AI-generated results.

05Learning sequence

1. Compare the curves and propose two explanations using concepts already studied.
2. Ask AI to translate each explanation into a predicted result for an additional controlled trial.
3. Choose one trial with the greatest power to separate the explanations and justify the choice.
4. Inspect the result and evaluate both the chemical explanation and possible measurement limitations.
5. Revise the explanation, predict another curve, and check it. Distinguish reaching the same extent sooner from producing a larger final amount.

07Evidence of learning

Annotated curves, a controlled experiment plan, and a revised explanation that connects particle-level reasoning with the measured trend.

01Curriculum connection

Balanced equations, stoichiometry, limiting reactants, theoretical yield, and mass conservation.

02Challenge

Can you detect and repair a production proposal whose quantities and promises conflict?

03Cognitive extension

Coordinate symbolic equations, quantitative constraints, and a changing design proposal. AI negotiates revisions as a fictional supplier; students use material balances to expose hidden assumptions and defend a feasible alternative.

04Preparation and AI role

Provide a teacher-checked reaction, molar masses, feed quantities, and fictional process costs. Include one deliberately inconsistent product claim. Define the system boundary and available measurements. Keep all work on paper or in a spreadsheet; no chemical production is required.

05Learning sequence

1. Balance the equation and calculate the limiting reactant and theoretical yield independently.
2. Interrogate AI’s supplier pitch. Request an account of excess reactants, byproducts, and any material crossing the system boundary.
3. Demonstrate the inconsistent claim using a calculation. Ask the supplier to revise one condition at a time.
4. Compare feasible revisions, considering reactant use, product purity, waste, and cost where data are supplied.
5. Defend a corrected proposal and explain what additional measurement is needed if reported product mass exceeds the theoretical yield.

07Evidence of learning

A balanced equation, material-balance calculation, revision log, and feasible recommendation with clearly stated assumptions.

01Curriculum connection

Game mechanics, scoring, state changes, loops, and playtesting.

02Challenge

What happens when the best way to win is different from the behavior you intended to encourage?

03Cognitive extension

Reason about incentives and emergent strategies across iterations. AI proposes adversarial player plans; students implement and test those plans in Unity to distinguish imagined exploits from reproducible design failures.

04Preparation and AI role

Build or reuse a small 3D collection game with a score, obstacles, and a restart condition. List the intended player behavior and explicit rules. AI receives the rule description and relevant logic; it has not played the build unless actual tool access is provided.

05Learning sequence

1. State the experience the game should encourage and predict how its scoring will produce that behavior.
2. Ask AI for contrasting strategies, including a plan that maximizes score while avoiding the intended challenge.
3. Attempt the proposed strategies in the actual build. Record which exploit works and the exact conditions required.
4. Revise one rule and request a new round of strategies that respond to the revision.
5. Retest with peers and defend the final trade-off between freedom, challenge, and preventing the exploit.

07Evidence of learning

A playable prototype, a reproducible exploit, test results before and after revision, and an explanation of the incentives behind both versions.

01Curriculum connection

Level layout, visual feedback, affordances, camera framing, and onboarding.

02Challenge

Can players discover a game mechanic without being told what to do?

03Cognitive extension

Model another person’s incomplete understanding and design evidence that changes it. AI generates competing interpretations of the same scene; students test those interpretations through observed human play rather than assuming their design communicates its intent.

04Preparation and AI role

Create a small 3D level that introduces one mechanic, such as activating a platform. Use screenshots, a scene description, and a list of visible cues. Recruit peers unfamiliar with the level for brief observation; AI predictions do not substitute for their behavior.

05Learning sequence

1. Remove explanatory text and predict the first action a new player will take, stating the cue behind that prediction.
2. Ask AI to infer several plausible but incompatible rules from the visible information alone.
3. Observe a new player without coaching. Record actions and ask afterward what they believed the rules were.
4. Redesign one cue or feedback signal to distinguish the intended rule from a demonstrated misunderstanding.
5. Test with another new player and explain whether improved understanding followed from clearer evidence or merely a simpler task.

07Evidence of learning

Before-and-after level versions, observed player actions, and an explanation linking a specific cue to a change in understanding.

01Curriculum connection

Camera perspective, information design, puzzle logic, and player communication.

02Challenge

How can two players interpret the same game state differently while both reason sensibly?

03Cognitive extension

Design and reason across unequal information states. AI generates contrasting mental models and challenges what each player could know; students build a puzzle that requires exchanging evidence rather than guessing a designer’s intention.

04Preparation and AI role

Use one scene with two selectable viewpoints or sequential roles; multiplayer networking is unnecessary. Give each role access to different clues about a common solution. Establish a shared solution key and verify that the clues are consistent.

05Learning sequence

1. Define the puzzle and map what each role can observe, infer, and mistakenly assume.
2. Give AI each role’s information separately. Ask it to propose a plausible interpretation without seeing the hidden solution.
3. Build the views and clues, then test whether each role has enough information to contribute something necessary.
4. Have two peers solve through communication. Record where their interpretations conflict or where one clue makes the other role unnecessary.
5. Revise the information distribution and explain how successful dialogue changes what the players can jointly know.

07Evidence of learning

A playable puzzle, an information map for both roles, and a playtest account showing how communication resolves a specific disagreement.

01Curriculum connection

Game state, telemetry, conditional rules, difficulty adjustment, and testing.

02Challenge

Can a game adapt to a player without mistaking exploration or experimentation for inability?

03Cognitive extension

Examine the limits of behavioral proxies and feedback-driven decisions. AI creates competing explanations for identical play data; students design, test, and revise a transparent adaptation rule while keeping player experience in view.

04Preparation and AI role

Use a small obstacle course and familiar scripting concepts. Record a few nonidentifying measures, such as attempts and completion time. Implement a simple adjustable parameter, such as obstacle speed. Use fictional traces first and voluntary peer testing afterward.

05Learning sequence

1. Define a difficulty-adjustment goal and code an initial rule based on selected measures.
2. Ask AI for two plausible player histories that produce similar measurements but reflect different intentions or needs.
3. Run those traces through the rule and identify decisions that are poorly supported by the available evidence.
4. Revise the design using a better signal, a player choice, or a reversible adjustment. Explain the trade-off.
5. Test the revised build and compare predicted benefit with player feedback. State what the data cannot establish about ability or enjoyment.

07Evidence of learning

An adaptive prototype, contrasting test traces, and a justification showing why the revised rule better matches the intended player experience.

01Curriculum connection

Heritable variation, natural selection, genetic drift, population change, and interpreting models.

02Challenge

Can the same change in a population arise through different evolutionary processes?

03Cognitive extension

Distinguish a plausible evolutionary story from a tested mechanism. AI sustains competing explanations across changing conditions; students use repeated simulations to examine chance, selection, and the limits of inference from an endpoint.

04Preparation and AI role

Use a teacher-checked population simulator with adjustable population size, selection, and random sampling. Supply several initial trajectories and a fixed menu of experiments. Numerical results must come from the simulator. Keep mutation and migration fixed or explicitly disabled.

05Learning sequence

1. Inspect a change in trait or allele frequency. Propose explanations involving selection and chance, identifying what the endpoint alone cannot establish.
2. Ask AI to defend each explanation from the available evidence and state what further pattern it predicts.
3. Design repeated trials that change one parameter, such as population size, while holding the others fixed. Predict the distribution of outcomes before running them.
4. Compare trajectories rather than a single run. Revise the relative support for the explanations and identify where both processes could contribute.
5. Test your explanation in a new condition and write a conclusion that avoids implying that organisms change because they need to.

07Evidence of learning

Competing explanations, graphs from repeated trials, and a defended conclusion explaining what the experiment reveals about process and what remains uncertain.

01Curriculum connection

Membranes, transport, cellular processes, and reasoning across molecular and cellular scales.

02Challenge

Can you identify why a cell changes when several mechanisms could produce a similar observation?

03Cognitive extension

Coordinate explanations across biological scales and design interventions that separate them. AI plays a panel of competing model builders; students choose evidence that connects a molecular proposal to a measurable cellular response.

04Preparation and AI role

Prepare a teacher-checked dataset or simulator for a familiar system, such as cell-volume responses to defined solutions. Provide membrane properties, solute behavior, measurement limits, and additional available observations. Restrict mechanisms to concepts already studied.

05Learning sequence

1. Describe the observed response without explaining it. Draw two candidate models showing particles, membrane properties, and relevant transfers.
2. Ask AI to develop the strongest version of each model from the supplied conditions, identifying assumptions rather than adding hidden facts.
3. Choose an intervention or additional measurement for which the models make different predictions. Explain the link between the molecular change and the cell-level outcome.
4. Examine the prepared result and revise the models. Check whether the evidence rules out a mechanism or only one version of it.
5. Predict a response under a new condition and state which aspect of the mechanism remains unobserved.

07Evidence of learning

Initial and revised diagrams, a discriminating investigation, and an explanation connecting molecular assumptions to observed cellular behavior.

01Curriculum connection

Food webs, population interactions, limiting factors, and ecosystem change.

02Challenge

Can an intervention solve an ecological problem while creating a larger one elsewhere?

03Cognitive extension

Reason through indirect effects, feedback, and uncertainty across a network. AI generates rival consequence pathways; students use ecological evidence and checked models to decide which pathways deserve attention and how to monitor an intervention.

04Preparation and AI role

Provide a bounded ecosystem case with population trends, a food web, and an explicit management goal. Use a teacher-checked model or staged evidence packet for intervention outcomes. Separate established relationships from uncertain ones; do not treat AI scenarios as forecasts.

05Learning sequence

1. Propose an intervention and map its intended direct effect alongside two possible indirect effects.
2. Ask AI to challenge the plan through a different pathway in the food web, identifying every assumption required.
3. Select an informative model run or new observation. Predict both an early effect and a delayed consequence before viewing results.
4. Revise the plan after examining the evidence. Test whether a second plausible model would lead to a different decision.
5. Recommend a staged intervention with a monitoring indicator and a clear condition for modifying or stopping it.

07Evidence of learning

A causal network, tested consequence pathways, and a recommendation with an evidence-based monitoring and revision plan.

01Curriculum connection

Inheritance, gene expression, environmental influence, variation, and causal inference.

02Challenge

What evidence would justify explaining a trait through genes, environment, or their interaction?

03Cognitive extension

Move beyond a single-cause account by comparing explanatory models. AI proposes competing interpretations of the same observations; students design comparisons that distinguish inherited differences from environmental effects and examine exceptions.

04Preparation and AI role

Use a teacher-curated dataset from a nonhuman system, such as plant growth across genotypes and environments. Include repeated observations and variation within groups. Define the experimental design and distinguish controlled data from observational data.

05Learning sequence

1. Describe the pattern and write an initial explanation before receiving AI’s alternatives.
2. Ask AI to construct genetic, environmental, and interaction-based accounts where the dataset permits them. Mark unsupported assumptions in each.
3. Choose a comparison or additional experiment that could distinguish the accounts. Explain which variables must be controlled and why replication matters.
4. Inspect an additional prepared dataset, including cases that challenge the initial explanation. Revise the account of group differences and individual variation.
5. Write a conditional prediction for a new environment and explain what the evidence cannot establish about any single organism.

07Evidence of learning

A comparison of competing models, an experimental design, and a revised prediction that distinguishes group patterns from individual certainty.

01Curriculum connection

Close reading, literary value, censorship arguments, and the consistency of evaluative principles.

02Challenge

Does your principle about access to a book survive a difficult case you did not choose?

03Cognitive extension

Formulate and test a general principle across unfamiliar cases. AI creates targeted challenges to the student’s own reasoning; students distinguish a principled qualification from an exception made to protect a preferred conclusion.

04Preparation and AI role

Choose a text studied in the course and prepare relevant passages with verified context about a documented challenge. Define the decision setting, such as a fictional school reading list. Distinguish required reading, optional access, and removal; this is an interpretive exercise, not a legal ruling.

05Learning sequence

1. State a position about the specified case and a principle that could apply beyond this particular book.
2. Ask AI to test the principle through hypothetical cases that vary one feature at a time, preserving the decision setting.
3. Explain your judgment in each case and identify what evidence about the actual text would be needed.
4. Return to the class text. Test whether your principle recognizes its form, context, and complexity rather than reacting to isolated content.
5. Write a revised principle and apply it to the original case and the strongest counterexample.

07Evidence of learning

A principle, a case-by-case reasoning log, and a revised judgment grounded in the literary text and its context.

01Curriculum connection

Narrative voice, irony, characterization, omission, and the relationship between form and meaning.

02Challenge

Can removing a troubling passage change the very idea a text invites readers to question?

03Cognitive extension

Trace the interpretive consequences of editorial intervention across a work. AI proposes contrasting edits and defends them; students examine how local wording changes affect ambiguity, moral judgment, and the reader’s understanding of the whole.

04Preparation and AI role

Select a short passage from an approved class text with sufficient surrounding context. Students supply the excerpt for analysis. Keep a clearly labeled original and hypothetical edited versions. AI may suggest local transformations, not fabricated versions of published editions.

05Learning sequence

1. Interpret the passage’s function in the work, distinguishing depiction, narrator position, and possible authorial purpose.
2. Ask AI for two hypothetical edits with different editorial aims, such as removing a phrase or adding explanatory framing.
3. Annotate what each edit preserves, removes, or unintentionally changes in voice, irony, and reader judgment.
4. Have AI defend the edit you consider most damaging. Respond using the passage and connections elsewhere in the work.
5. Recommend retention, framing, or an alternative intervention, explaining the literary cost and limits of your choice.

07Evidence of learning

Annotated original and edited excerpts, with a text-supported explanation of how an editorial choice changes interpretation.

01Curriculum connection

Reader response, ambiguity, rhetoric, and claims about the effects of literature.

02Challenge

How can you evaluate claims about what a book does to readers when responses differ?

03Cognitive extension

Separate textual possibilities, predictions about reception, and actual evidence of response. AI constructs contrasting reader interpretations; students test the assumptions behind claims of harm, insight, or persuasion without treating simulated readers as real audiences.

04Preparation and AI role

Choose a passage and provide contextual information plus a small packet of authentic, contrasting reviews or reception documents. AI reader profiles should differ in stated knowledge, interpretive assumptions, or reading purpose, without claiming to represent demographic groups.

05Learning sequence

1. Write two possible effects of the passage and identify the textual feature and reader assumption behind each prediction.
2. Ask AI to construct contrasting readings from clearly defined interpretive starting points.
3. Interrogate each reading: what did it notice, assume, overlook, or import from outside the text?
4. Compare the hypothetical readings with the authentic reception packet. Identify agreement, disagreement, and the limits of a small selection of responses.
5. Rewrite a broad claim about the book’s effects as a conditional argument that identifies what further evidence would be necessary.

07Evidence of learning

A map linking textual features to possible responses, an analysis of real reception evidence, and a carefully qualified claim about effects.

01Curriculum connection

Literary access, canon formation, classification, selection, and visibility.

02Challenge

Can a collection restrict what readers encounter even when every title remains available?

03Cognitive extension

Investigate how selection rules and categories shape access without collapsing different forms of restriction into one label. AI generates recommendation orders under explicit rules; students audit how those rules distribute visibility and alter the apparent literary landscape.

04Preparation and AI role

Prepare a small catalog of course texts or verified summaries, with stable IDs and teacher-specified metadata. Include at least one text read closely. Create a fictional discovery interface that shows only a few recommendations; no software build or real platform data is required.

05Learning sequence

1. Define what a useful recommendation should achieve and select a transparent ranking rule.
2. Ask AI to apply that rule to the catalog. Verify the ranking manually and inspect which books repeatedly disappear from the visible shortlist.
3. Compare a second rule, such as variety of themes or familiarity, using the same catalog and display limit.
4. Use the closely read text to challenge how a category or summary misrepresents its complexity. Revise the metadata or rule and test again.
5. Explain the differences among removal, selection, low visibility, and misclassification, then recommend a discoverability policy with a stated trade-off.

07Evidence of learning

Two verified recommendation lists, a literary critique of one classification, and a policy argument explaining who or what the design makes easier to encounter.

01Curriculum connection

Formal analysis, iconography, historical context, and the interpretation of visual evidence.

02Challenge

How does contextual information change what you believe you can see in an artwork?

03Cognitive extension

Distinguish visual observation from contextual inference across successive interpretations. AI builds rival readings from limited information; students test how new evidence changes confidence without allowing a convincing label to substitute for looking.

04Preparation and AI role

Choose a work from the current unit and prepare a high-quality reproduction plus a verified context packet. Withhold identifying metadata initially. AI may recognize the work despite the prompt, so students must flag any outside knowledge it introduces.

05Learning sequence

1. Describe visible features before proposing meaning. Separate what is observed from what is inferred.
2. Ask AI for two contrasting interpretations grounded only in the supplied image or detailed description.
3. Identify the visual feature each reading explains best and the feature it struggles to explain.
4. Reveal verified context, such as original setting, patronage, or audience. Reassess the interpretations and return to the image to test the revised claims.
5. Write a final analysis distinguishing visible evidence, documented context, and remaining uncertainty about meaning or intention.

07Evidence of learning

An observation record, two tested interpretations, and a final analysis linking visual details to documented context.

01Curriculum connection

Patronage, materials, function, audience, site, and artistic choices.

02Challenge

Which features of an artwork would change if its patron, setting, or purpose changed?

03Cognitive extension

Reason backward from historical constraints to artistic decisions, then transfer that reasoning to a counterfactual commission. AI plays a demanding fictional patron; students defend design choices through comparison with a documented work.

04Preparation and AI role

Select a work with reliable information about material, site, function, and commission. Provide reproductions and a concise source packet. Students produce sketches or an annotated design brief; the hypothetical redesign must remain distinct from the historical artwork.

05Learning sequence

1. Analyze how the documented work responds to its material, audience, setting, and purpose.
2. Change one commission condition, such as scale, viewing location, or intended audience, while holding the historical period and other constraints fixed.
3. Ask AI to act as a fictional patron pressing for changes within the revised brief. Evaluate requests against the packet’s technical and historical constraints.
4. Sketch a redesign and defend what you retain, transform, or reject. Identify an artistic effect that cannot transfer unchanged.
5. Return to the original and explain what the redesign experiment reveals about a choice previously treated as merely stylistic.

07Evidence of learning

An annotated redesign with source-based constraints and a comparative analysis showing how function and setting shape artistic form.

01Curriculum connection

Curatorial interpretation, comparison, historical relationships, and canon formation.

02Challenge

How can the same group of artworks tell substantially different histories of art?

03Cognitive extension

Construct and test historical narratives through selection, sequence, and juxtaposition. AI proposes rival exhibition logics; students examine whose work becomes central, what connections are supported, and which narrative each arrangement obscures.

04Preparation and AI role

Prepare a fixed set of six to eight works from the course with reproductions and verified metadata. Include enough context to support more than one organizing principle. Students can use paper layouts or slides; no actual exhibition installation is required.

05Learning sequence

1. Arrange the works into an exhibition and state its central historical argument.
2. Ask AI for a rival arrangement of the same works using a different principle, such as materials, exchange, function, or patronage.
3. Compare how the arrangements distribute centrality and imply relationships. Verify any proposed connection against the source packet.
4. Write a short label for one juxtaposition in each arrangement, grounding the claim in visual details and documented context.
5. Revise the exhibition and acknowledge an important history that remains difficult to tell with the available works.

07Evidence of learning

Two exhibition plans, comparative object labels, and a defended curatorial argument that identifies the limits of the selected collection.

01Curriculum connection

Material history, condition, conservation, authenticity, and uncertainty in reconstruction.

02Challenge

When does making an artwork easier to understand risk making it less faithful to the surviving evidence?

03Cognitive extension

Evaluate interventions under competing definitions of authenticity. AI sustains contrasting conservation arguments; students weigh visual legibility, material evidence, uncertainty, and reversibility in a documented or clearly fictional decision case.

04Preparation and AI role

Select a teacher-prepared case with a reproduction, condition map, and verified information about an artwork’s changes. Specify known losses, later additions, and unknowns. Proposals remain on paper or digital mockups; students do not intervene in an original object.

05Learning sequence

1. Distinguish surviving material, documented change, and uncertain reconstruction. Record an initial intervention proposal.
2. Ask AI to defend two alternatives, such as minimal stabilization and a more legible presentation, using only the case evidence.
3. Question what each proposal preserves, removes, or requires viewers to infer. Identify assumptions about the intended audience and original appearance.
4. Develop an annotated proposal showing where uncertainty would remain visible and how reconstructed elements would be disclosed.
5. Defend the proposal against its strongest objection and state what new evidence would lead you to reconsider it.

07Evidence of learning

An annotated intervention proposal and a justification comparing visual understanding, material evidence, uncertainty, and reversibility.

01Curriculum connection

Memoir, scene construction, reflective voice, and the distinction between recollection and documentation.

02Challenge

How can you write a vivid scene when the evidence complicates what you remember?

03Cognitive extension

Hold several possible accounts in mind while deciding what can responsibly be asserted. AI generates rival interpretations of a remembered event; students verify details and make uncertainty part of the literary form rather than filling gaps with persuasive invention.

04Preparation and AI role

Choose an ordinary, non-sensitive event suitable for sharing, or use a teacher-supplied autobiographical account. Gather available notes, photographs, or other records. AI can question a memory but cannot recover it or establish another person’s thoughts.

05Learning sequence

1. Write a short scene from memory before consulting supporting records. Mark remembered dialogue and uncertain details.
2. Compare the draft with available evidence and build a ledger of corroborated, remembered, inferred, and unknown details.
3. Ask AI for two possible interpretations of the discrepancies without adding events. Identify what evidence would distinguish them.
4. Revise the scene yourself, choosing where to qualify, omit, or reflect on an uncertain detail while preserving narrative movement.
5. Write an author’s note explaining a significant choice about factual confidence and how it affects the reader’s trust.

07Evidence of learning

An initial scene, an evidence ledger, a revised scene, and an author’s note demonstrating how factual uncertainty shaped a literary decision.

01Curriculum connection

Narrative structure, chronology, emphasis, juxtaposition, and implied causation.

02Challenge

Can a nonfiction piece mislead without containing a single false sentence?

03Cognitive extension

Trace how arrangement produces inferences beyond stated facts. AI constructs contrasting structures from a fixed record; students examine how sequencing and omission shape blame, significance, and the appearance of causality.

04Preparation and AI role

Prepare a short verified factual record about an ordinary event, with source IDs and a reliable chronology. Include enough context to support different emphases. Students produce their own final narrative; AI provides outlines rather than a finished essay.

05Learning sequence

1. Write a brief account and identify the inference you expect readers to make about the event.
2. Ask AI for two outlines using the same facts but encouraging contrasting interpretations. Require a record of every omitted fact.
3. Audit both outlines for implied chronology, causation, motives, and certainty. Identify an implication that exceeds the evidence.
4. Select a structure and write the piece yourself, adding context or changing sequence where necessary to make the implied claims defensible.
5. Ask a peer to state what they inferred, compare that response with your prediction, and explain the revision it prompts.

07Evidence of learning

Two annotated outlines and a student-written piece with a commentary linking structural choices to supported and unsupported reader inferences.

01Curriculum connection

Braided essays, scene and reflection, research integration, scale, and representativeness.

02Challenge

How can a personal story illuminate a wider issue without pretending to represent everyone?

03Cognitive extension

Move between individual experience and population-level evidence without collapsing either scale. AI exposes tensions between a scene and a research claim; students develop a structure that preserves both human specificity and the limits of generalization.

04Preparation and AI role

Pair an ordinary personal scene or teacher-supplied narrative with a concise, verified research packet on the same theme, such as commuting or public space. Identify populations, dates, definitions, and limitations in the research. No personal disclosure is required.

05Learning sequence

1. Write the scene and separately summarize the research finding, stating what each can establish.
2. Ask AI to locate points of connection and tension, including where the experience may be an exception rather than an illustration.
3. Design two possible braids between scene and research. Predict what each transition invites readers to infer.
4. Write the essay yourself, using transitions to distinguish lived meaning, evidence, and interpretation rather than forcing agreement.
5. Audit the draft with AI for unsupported movement from “I” to “we,” then revise and explain one place where the strands remain unresolved.

07Evidence of learning

A braided essay with source references and an annotated transition showing how the writing connects scales without overgeneralizing.

01Curriculum connection

Character portrayal, narrator authority, ethical representation, omission, and reflective writing.

02Challenge

How can you portray another person vividly without claiming knowledge you do not possess?

03Cognitive extension

Examine the narrator’s power through sustained challenges to framing and attribution. AI acts as an editorial critic of representation; students distinguish an observed action from an assigned motive and defend what belongs in the story.

04Preparation and AI role

Use a teacher-supplied nonfiction vignette or a student draft with identifying details removed. Select material suitable for classroom discussion. AI-generated objections are possibilities to consider, not testimony or consent from the person portrayed.

05Learning sequence

1. Annotate a scene for observable behavior, attributed motives, private information, and evaluative language.
2. Ask AI to raise the strongest possible objections to how the narrator frames the other person, without pretending to know that person’s response.
3. Examine which objections concern accuracy, relevance, exposure, or a disagreement the narrator can reasonably retain.
4. Write two versions that handle one difficult passage differently. Compare what each gains and loses in clarity, complexity, and fairness.
5. Choose and defend a final version, explaining where consultation or additional evidence would be needed before any real publication.

07Evidence of learning

An annotated scene, two alternative treatments of a passage, and a defended revision explaining the narrator’s evidence and responsibilities.

01Curriculum connection

Distributive justice, impartiality, equality, need, and justification of principles.

02Challenge

Would you endorse the same allocation rule before knowing whether it benefits you?

03Cognitive extension

Reason across possible social positions and test the consistency of a principle. AI supplies contrasting positions and difficult cases; students examine how impartial reasoning interacts with needs that a uniform rule may overlook.

04Preparation and AI role

Use a fictional allocation problem, such as access to limited rehearsal space. Supply resource limits and relevant differences in need. Prepare concealed role cards. Ground discussion in course readings rather than treating any AI formulation as a complete theory of justice.

05Learning sequence

1. Choose an allocation rule before receiving a role. Explain which differences should matter and which should not.
2. Ask AI to test the rule against contrasting fictional positions and identify its strongest justification and objection.
3. Reveal role cards and apply the rule. Record whether your judgment changes and distinguish self-interest from newly relevant information.
4. Compare an alternative principle from the course readings. Identify the burden each rule leaves someone carrying.
5. Defend a final rule, including a justified exception or appeal process where needed, and explain whether you could endorse it from an unfavorable position.

07Evidence of learning

An initial rule, an application across roles, and a revised justification that addresses impartiality, relevant differences, and the strongest objection.

01Curriculum connection

Consequences, duties, care, moral conflict, and the limits of compromise.

02Challenge

How should you act when every available option leaves a serious moral claim unmet?

03Cognitive extension

Sustain several ethical frameworks without reducing them to slogans or a simple vote. AI develops competing arguments from course readings; students decide whether disagreement concerns facts, values, obligations, or the framing of the available choices.

04Preparation and AI role

Prepare a bounded fictional case, such as honoring an agreed use of a community venue when an urgent competing need arises. Specify alternatives and uncertain consequences. Supply short extracts from ethical approaches already studied; AI must work from these accounts.

05Learning sequence

1. Make an initial decision and name the moral claim you find hardest to satisfy.
2. Ask AI to develop contrasting arguments through the supplied approaches. Reconstruct each argument before criticizing it.
3. Separate disagreements about likely consequences from disagreements about what ought to matter. Identify which could be resolved by more evidence.
4. Challenge the option set and consider a feasible alternative. Test whether it resolves the conflict or relocates its cost.
5. Defend an action and state the remaining obligation, such as explanation, repair, or reconsideration, rather than claiming the decision eliminates every moral loss.

07Evidence of learning

An argument comparison and a decision statement that distinguishes empirical uncertainty, normative disagreement, and the obligations remaining after action.

01Curriculum connection

Moral agency, foreseeability, control, institutional responsibility, and accountability.

02Challenge

How should responsibility be assigned when harm emerges from many individually limited decisions?

03Cognitive extension

Map responsibility across a system while distinguishing causing an outcome from having duties to prevent or repair it. AI constructs role-based defenses; students test those defenses against information, authority, benefit, and opportunities to intervene.

04Preparation and AI role

Use a fictional case in which an automated booking system repeatedly disadvantages some users. Provide a timeline of decisions by designers, managers, operators, and reviewers. Specify what each knew and controlled. The activity concerns moral responsibility, not a legal liability determination.

05Learning sequence

1. Map the decisions and their contributions to the outcome. Record who could know, change, or challenge each stage.
2. Ask AI to present the strongest defense for each role using only the case facts.
3. Interrogate the defenses, separating limited control from a complete absence of responsibility. Identify a point where intervention was possible.
4. Compare duties before the harm, responsibility for the outcome, and obligations to repair it. Avoid assuming these must fall on the same actor.
5. Propose an accountability arrangement with a named decision owner, meaningful review, and a way for affected people to challenge an outcome.

07Evidence of learning

A responsibility map and an accountability proposal supported by evidence about control, knowledge, intervention opportunities, and repair.

01Curriculum connection

Procedural justice, incentives, strategic behavior, and the relationship between principles and institutions.

02Challenge

Can a fair rule remain fair when people adapt to it?

03Cognitive extension

Test ethical principles within dynamic systems rather than isolated dilemmas. AI creates successive plausible adaptations to a policy; students assess how revisions affect incentives, access, dignity, and the burdens of enforcement.

04Preparation and AI role

Create a fictional policy for distributing a limited shared resource, such as community workshop hours. Specify the goal, available information, and administrative limits. Use course readings on justice to justify criteria. AI scenarios are possible behaviors, not evidence about actual groups.

05Learning sequence

1. Design a rule and explain its ethical aim, including who may be disadvantaged by its operation.
2. Ask AI for three possible responses: ordinary use, strategic exploitation, and difficulty complying despite legitimate need.
3. Apply the policy to each case. Revise it and ask AI how behavior might change in response.
4. Examine whether preventing exploitation adds disproportionate barriers or monitoring burdens. Compare a simpler alternative.
5. Defend a final rule with an appeal mechanism and a review trigger, explaining how you would recognize failure without assuming every user acts strategically.

07Evidence of learning

A policy revision log and a defended rule connecting its ethical principle to incentives, access, appeals, and conditions for review.

01Curriculum connection

Life cycle reasoning, functional units, resource use, emissions, and environmental trade-offs.

02Challenge

Can two comparisons reach opposite conclusions about the same products without either calculation being wrong?

03Cognitive extension

Coordinate system boundaries, repeated use, and different impact measures. AI challenges the assumptions behind an environmental comparison; students identify the conditions under which a recommendation holds and where one impact cannot represent the whole picture.

04Preparation and AI role

Provide a teacher-checked dataset for a familiar choice, such as reusable and single-use containers. Include production, transport, use, and end-of-life assumptions with units and ranges. Define a common service, such as delivering the same number of drinks. Do not use AI-generated quantities as evidence.

05Learning sequence

1. Define the functional unit and system boundary, then make an initial comparison using the supplied data.
2. Ask AI to challenge an omitted stage or an assumption about reuse, washing, loss, or transport.
3. Calculate a break-even point where the data support one, and test how it shifts when a key assumption changes.
4. Compare another supplied impact measure. Identify whether the preferred option changes or whether the measures cannot responsibly be collapsed into one score.
5. Write a conditional recommendation specifying the operating conditions, remaining trade-off, and measurement most likely to change it.

07Evidence of learning

A transparent comparison with units, a sensitivity or break-even analysis, and a recommendation stating the conditions under which it is supported.

01Curriculum connection

Environmental systems, scenario analysis, adaptation, uncertainty, and monitoring.

02Challenge

How can you choose an environmental plan when you cannot confidently predict which future will occur?

03Cognitive extension

Compare decisions across plausible scenarios rather than optimize for one expected future. AI proposes coherent stress tests and challenges hidden assumptions; students seek an adaptable plan and specify evidence that would trigger a change.

04Preparation and AI role

Use a bounded fictional case, such as reducing heat exposure in a campus courtyard. Supply checked data for options and several plausible environmental conditions, with an explicit budget and decision criteria. Scenario ranges are stipulated assumptions, not forecasts or assigned probabilities.

05Learning sequence

1. Choose a plan for the initial conditions and state what you are assuming about future use and environmental change.
2. Ask AI to challenge the plan under contrasting scenarios that remain within the supplied ranges.
3. Compare the options across scenarios, identifying where each fails rather than merely calculating an average score.
4. Design a staged or reversible plan that preserves useful choices. Explain what performance or cost you sacrifice to remain adaptable.
5. Recommend the plan with a monitoring indicator and a threshold for reconsideration, distinguishing evidence-based triggers from value choices.

07Evidence of learning

A scenario comparison and an adaptive recommendation with explicit failure conditions, monitoring indicators, and revision triggers.

01Curriculum connection

Environmental monitoring, causal inference, baseline selection, confounding, and uncertainty.

02Challenge

Does an improved environmental indicator show that a particular intervention caused the improvement?

03Cognitive extension

Test a success claim against alternative causal explanations. AI constructs rival accounts and scrutinizes the evaluation design; students use time patterns, comparisons, and mechanism evidence to judge how much credit an intervention deserves.

04Preparation and AI role

Prepare a checked dataset around a fictional or documented intervention, such as a campus energy-saving program. Include dates, relevant weather or occupancy information, and a comparison where available. Clearly label simulated datasets. Specify measurement uncertainty and missing information.

05Learning sequence

1. Graph the indicator before and after the intervention and write an initial success claim.
2. Ask AI for alternative explanations consistent with the available data, such as changing weather, occupancy, or an existing trend.
3. Choose a fairer comparison or adjustment and predict how it might change the interpretation. Check comparability rather than assuming a control is equivalent.
4. Reanalyze the supplied data and examine whether the proposed mechanism leaves an observable trace.
5. Rewrite the claim to match the strength of the evidence and propose the most informative next measurement or evaluation change.

07Evidence of learning

Before-and-after and revised analyses, an explicit treatment of an alternative explanation, and a calibrated conclusion with a next-step evaluation plan.

01Curriculum connection

Spatial environmental data, exposure, distribution, environmental justice, and intervention priorities.

02Challenge

Can the option with the best average environmental outcome be a poor choice for the most affected people?

03Cognitive extension

Integrate environmental measurement with explicit distributional judgments. AI challenges plans from contrasting stated priorities; students distinguish disagreements about data from disagreements about which burdens should receive attention.

04Preparation and AI role

Provide a fictional neighborhood dataset for an issue such as heat exposure, with population, exposure patterns, intervention effects, costs, and uncertainty. Use maps or a compact spatial table. Clarify whether measures describe ambient conditions or personal exposure; avoid unsupported health-risk calculations.

05Learning sequence

1. Use the budget to select a plan that maximizes an explicitly defined aggregate environmental benefit.
2. Map who benefits and who remains highly exposed. Examine population weighting and variation within each area.
3. Ask AI to challenge the plan from different priorities, such as reducing the highest exposure or achieving broad coverage, without pretending to speak for real communities.
4. Construct a second feasible allocation and test both against uncertain effectiveness estimates and the chosen distributional criteria.
5. Recommend a plan, separating scientific estimates from ethical priorities and identifying whose knowledge would be needed before a real decision.

07Evidence of learning

Two feasible allocations, a map or table showing benefits and remaining burdens, and a recommendation explaining evidence, uncertainty, and value priorities.