AI and Civic Reasoning | Weighing Evidence and Values in Shared Decisions

A civic question concerns how people live together and organise something they share. A room timetable looks modest, but it raises recognisable questions about access, competing purposes, responsibility and the explanation owed to affected users. Students can practise these questions without pretending to settle a national controversy or speak for a real community.

The fictional Willow Learning Room offers quiet individual work and collaborative projects. Both are legitimate uses of the room. The planning question is: how should its six weekly sessions be arranged for the next four-week cycle, given the recorded use and the interests described in the pack? This wording leaves the answer open. “How do we stop projects wasting space?” would already contain a disputed conclusion.

Ask learners to write the question in their own words before asking AI for a solution. They should identify what is shared, what is limited and why more than one person has a stake. A learner who thinks the task is simply to maximise attendance may otherwise answer confidently without noticing that another participant values dependable weekly access.

The article on society and collective intelligence connects public learning with institutions. This enquiry brings that connection into a manageable classroom discussion: several people contribute knowledge, but the group still needs to explain how those contributions affect its recommendation.

Contents · Next chapter

The statement “Monday was full in all four recorded weeks” can be checked against a register. “An additional Thursday quiet session would fill” is a prediction. “Maintaining a dependable weekly opportunity matters even when a group is small” expresses a priority. “Try alternating Thursday activities” proposes an action. These sentences do different jobs, even when they appear together in one persuasive paragraph.

The Council of Europe’s model of competences for democratic culture, incorporated in its 2018 Reference Framework, organises twenty competences across values, attitudes, skills, and knowledge and critical understanding. It offers a conceptual framework for educational use. It is not an AI intervention study, an endorsement of this room lesson or a Singapore assessment rule.

In the classroom case, a value becomes useful to discussion when the learner explains what it would require. “Fairness” might mean equal session numbers, attention to unmet need, protection of a regular opportunity, or a transparent chance to challenge the decision. Those interpretations can lead to different arrangements. Naming the word alone does not settle which interpretation should guide this choice.

Invite students to classify one sentence from their first response, then explain the classification. Some sentences combine categories. “We should add a quiet session because it will attract more users” joins a recommendation to a prediction. The group can accept the importance of access while asking for better support for the forecast.

Contents · Previous chapter · Next chapter

The room has twenty usable places per session. Every listed session lasts two hours. In this exercise, staffing, room suitability and operating cost are held equal across the listed activity types. Both activities are assumed suitable for the room. These are simplifying conditions supplied to make the timetable question tractable, not findings about real service costs or suitability.

There are six fixed slots each week: Monday, Tuesday, Wednesday and Thursday from 4–6 pm, Saturday from 10 am–noon, and Saturday from 1–3 pm. No additional room or opening period is available in the initial case. The group may recommend changing Thursday’s activity. All other slots retain their listed activities. A four-week trial is possible, followed by a review before another cycle is chosen.

The fictional room committee has decision authority. Students act as a learning panel that offers advice. The committee is expected, within the scenario, to explain its response to that advice. Students do not issue bookings, promise a change to actual users or treat their classroom vote as a real public mandate.

Keep three tasks distinct: understand the existing arrangement, recommend a feasible alternative, and decide what evidence to gather next. A question such as “Could a second room solve this?” may be useful for future planning, but it cannot silently remove the current six-slot constraint. An ambitious possibility and an immediately available option belong in different parts of the discussion.

Contents · Previous chapter · Next chapter

Record A is the fictional room register for four consecutive ordinary teaching weeks. Each cell counts attendances at that session: one person attending once contributes one attendance. Someone returning on another day contributes again. The register supplies no identities, waiting-list record, reasons for absence or attendance outside these slots. All twenty places are available in every recorded session.

The repeated numbers deliberately simplify the arithmetic. They are not presented as a realistic forecast of weekly variation. Quiet work records 54 attendances each week and 216 across four weeks. Projects record 36 each week and 144 across four weeks. The combined total is 360 attendances across twenty-four sessions.

Reading the rows matters as much as adding them. Monday and Wednesday quiet sessions are full; Saturday quiet sessions have six unused places. Saturday projects are full, while Tuesday and Thursday projects each have twelve unused places. An activity-level average would conceal these differences between times. Learners should point to a particular slot when explaining an access problem.

Contents · Previous chapter · Next chapter

Quiet work accounts for 216 of the 360 recorded attendances, or 60%. Projects account for 40%. These are shares of recorded use under this timetable. They do not show that 60% of unique residents prefer quiet work, that the community voted for it, or that the two activities serve separate groups of people.

Each activity had twelve sessions and 240 available place-visits across the four weeks. A place-visit means one available place at one session. The room therefore offered 480 place-visits altogether. Quiet occupancy was 216 ÷ 240 = 90%; project occupancy was 144 ÷ 240 = 60%. The activities had equal nominal capacity but different recorded occupancy.

Neither number reveals how many people wanted to attend but could not. A full Monday register establishes that all places were used; it does not establish the size of a queue beyond twenty. An underfilled Thursday project register does not establish that its eight attendees could move to Tuesday or Saturday. Their availability is a separate question.

Ask the class to repair this invented claim: “Sixty per cent of the community wants quiet work, so four sessions must be quiet.” A sound correction preserves the useful 60% calculation, restores its attendance denominator and identifies the missing step from past use to a future allocation. The learner is explaining the evidence and the decision rule, not simply replacing a misleading phrase with a cautious one.

Contents · Previous chapter · Next chapter

Record B contains three invented comments offered voluntarily at an open discussion. Mira says, “Monday is full when I use the quiet session. I could also come on Thursday.” Dev says, “Thursday is the only one of the three project times I can attend.” Neri says, “I use both activities. A changing timetable could be confusing, although a calendar published before the cycle would help.” These comments are the complete comment set supplied for this exercise.

Mira provides a possible reason to test Thursday quiet work. Her statement does not establish how many others would attend. Dev identifies a potential loss that the occupancy totals cannot reveal. His statement does not establish that all eight Thursday attendances each week came from people with the same restriction. Neri challenges the assumption that quiet and project users form separate camps.

Do not count the comments as three votes between fixed proposals. None was collected through a representative recruitment process, and the speakers were not asked the same preference question. A comment can be relevant without being a measure of how widespread its concern is.

A useful follow-up asks about a practical condition: which sessions can you attend, what would changing Thursday prevent, and would a predictable alternating calendar work for you? People should have a way to correct the panel’s paraphrase. In a real enquiry, participation arrangements would need to suit the actual users; these fictional voices supply a starting point for classroom reasoning.

Contents · Previous chapter · Next chapter

Record C is the fictional coordinator’s operations note. It confirms the six slots, twenty places, equal operating assumptions, Thursday as the only changeable activity slot, and the possibility of a four-week trial. The room cannot host both activities simultaneously. Record D is an invented social post stating, “Project sessions are mostly empty; converting Thursday will solve the queue.” A second invented post copies the same sentence and names Record D as its source. Neither post contains additional observations.

Students should identify where each part of Record D came from. The project occupancy calculation could come from Record A, but the Saturday project session is full. The predicted solution to a queue has no queue record behind it. Two copies of one assertion provide two places where it appears, not two independent measurements.

Wineburg and McGrew’s 2019 study of lateral reading compared ten historians, ten professional fact checkers and twenty-five university students evaluating online information. The reported fact-checker approach included leaving a site to investigate it through other sources. This supports considering evidence beyond a page’s own presentation. The study did not test this lesson or establish AI-assisted civic learning effects.

For the present pack, tracing means returning to the named records and their scope. In a live web enquiry, it would also involve checking who produced a claim and what independent information supports it. The news and verified understanding article develops that wider information question.

Contents · Previous chapter · Next chapter

The learning panel now proposes three criteria: respond to demonstrated use, retain a meaningful opportunity for both activities, and avoid removing a known time-dependent opportunity without addressing the loss. These are the panel’s proposed priorities. They are not facts hidden in the attendance register or official rules supplied by an external framework.

For a concrete minimum, the panel asks that each activity retain at least two sessions every week. This is an original classroom decision constraint. It protects some continuing provision but does not guarantee that every existing user can attend the remaining times. Dev’s concern therefore survives even if a proposal passes the numerical minimum.

Invite another group to challenge the criteria before seeing which option they favour. Why protect two sessions rather than one? Should a crowded Monday matter more than a lightly used Thursday? How should uncertainty about future attendance affect the willingness to run a trial? These questions expose the reasoning that would otherwise disappear inside a final recommendation.

A score out of ten for each option would require an explanation of the scale and weights. If a student assigns “fairness: eight”, ask what observable consequence the number summarises. In this task, a short comparison of gains, losses and unknowns is more inspectable than an unexplained total. Students can still make a clear decision; they simply make its priorities available for discussion.

Contents · Previous chapter · Next chapter

The initial alternatives are A, keep Thursday for projects every week, and B, change Thursday to quiet work every week. After learners identify the constraint and inspect the records, the teacher can offer this constructed AI suggestion: “Could Thursday alternate between projects and quiet work across the four-week cycle, with the calendar announced in advance?” The suggestion is invented for the lesson; it is not a claim about a particular product’s observed response.

This becomes Option C: Thursday projects in Weeks 1 and 3, Thursday quiet work in Weeks 2 and 4. All other sessions remain unchanged. The suggestion creates a third arrangement that learners can examine. It adds no room hours, preserves the twenty-place limit, and retains at least two weekly sessions for each activity.

Students must develop the idea beyond its appealing label. Alternation leaves Dev with a Thursday project opportunity in only two of four weeks. A published calendar addresses Neri’s planning concern in principle, but it does not prove that everyone will find the arrangement convenient. Mira’s availability makes Thursday quiet use plausible without predicting attendance at either trial session.

Ask AI for alternatives within the supplied conditions, rather than for an authoritative declaration of fairness. Learners should keep the original question visible and record which suggestion changed their thinking. A proposal that violates a constraint can become a future enquiry, but it should not compete as if it were immediately available.

Contents · Previous chapter · Next chapter

Count sessions over the same four-week period. A has twelve quiet and twelve project sessions. B converts four Thursday sessions, giving sixteen quiet and eight project sessions. C converts two Thursdays, giving fourteen quiet and ten project sessions. Each arrangement still contains twenty-four sessions and forty-eight room hours.

C is not automatically best because its numbers sit between A and B. A compromise can still cause a serious loss or fail to address the main problem. The table makes the arrangement explicit; the class must judge whether its consequences fit the stated priorities.

Nor can the group subtract past project attendances from future capacity and declare that everyone fits. Under B, 160 project place-visits exceed the historical 144 attendances, yet Dev cannot use the remaining project times. Aggregate spare capacity and usable opportunities are different. Ask learners to connect at least one table entry to an actual condition in the comment cards.

Contents · Previous chapter · Next chapter

The OECD’s 2020 principles for deliberative processes connect a clearly defined public question with relevant evidence, inclusive participation, careful discussion and an accountable response. They address representative public deliberation. A classroom group using invented records is not such a representative process, and this lesson does not claim to reproduce or validate one.

In the original classroom discussion, one learner argues for B because the weekday quiet sessions are full. A second argues for A because a timetable should remain dependable for people already using it. Before replying, each learner explains the other’s strongest reason. This is an opportunity for correction: the second learner may value continuity without claiming that an existing arrangement must never change.

A useful exchange might be: “You want more quiet provision because Monday and Wednesday have no spare places. I accept that observation. I am asking whether Thursday is usable by the people who need another quiet time, and what happens to Thursday project users.” The reply can then address those questions instead of defending against an accusation of indifference.

The teacher can distinguish a challenge to a claim from a judgement about the speaker. “The register does not count unique people” identifies a problem that can be checked. “You do not care about the community” attributes a motive that the pack does not establish. Students need room to disagree and revise their position without the exercise becoming a performance of personal loyalty.

Contents · Previous chapter · Next chapter

A bounded AI task can help after the first discussion: organise each proposal under what it preserves, what it changes, what evidence supports it and what remains unknown. Give the tool the same records available to the learners. The group then checks each entry against those records and supplies its own judgement about priorities.

For C, a useful constructed summary would say that it preserves two Thursday project sessions, introduces two Thursday quiet sessions, and requires a predictable calendar. It would connect Mira’s comment to a possible quiet use, Dev’s comment to a remaining loss and Neri’s comment to planning. This organisation can make relationships easier to discuss without claiming that the three speakers endorsed the option.

By contrast, “All users agree that alternating is fair” has no support in Record B. Even a summary that lists every concern can mislead if its final sentence converts consideration into consent. Ask learners to mark the precise sentence where the evidence stops and an invented agreement begins.

AI can also suggest a question the class overlooked: “Would a fortnightly project session still allow a continuing project to progress?” The pack does not answer it. That is a useful contribution because it identifies a specific condition to investigate. The next step is to ask affected participants, not to have the model manufacture their reply. Keep the suggested question separate from the actual comment record.

Contents · Previous chapter · Next chapter

The following is an original model response, not the only acceptable answer. It recommends a conditional trial of C and makes the cost of that choice visible. Learners should compare its reasoning with their own priorities rather than copying its final preference.

We recommend considering Option C for one four-week cycle, after the committee checks the practical effect with Thursday project users. Thursday would remain a project session in Weeks 1 and 3 and become quiet work in Weeks 2 and 4. Every other session would keep its current activity.

Record A shows full Monday and Wednesday quiet sessions, while Thursday projects recorded eight attendances each week. This makes a limited test of Thursday quiet use worth considering. It does not show how many people would attend the new sessions or how many unique people use either activity.

The proposal retains at least two weekly sessions for each activity and creates two opportunities to observe Thursday quiet use. Its strongest objection is the loss of weekly Thursday projects. Dev cannot use the other project times; two retained Thursdays do not remove that loss. Neri’s request for a clear calendar also matters.

Before changing the calendar, the committee should establish whether affected users face a loss the proposed arrangement cannot adequately address. If that remains unresolved, keeping A for the next cycle is reasonable while another arrangement is investigated. If C proceeds, publish the whole cycle in advance and review attendance by slot, difficulties using the remaining times and participants’ accounts of what changed.

The committee should explain its decision and response to the objection. Two trial quiet sessions can inform another decision, but cannot establish a lasting demand pattern or a general improvement in learning.

The response connects evidence, priorities, conditions and review. A well-supported recommendation for A or B could also demonstrate strong reasoning. Assessment should examine those connections, not whether the learner chose the teacher’s preferred timetable.

Contents · Previous chapter · Next chapter

A recommendation is incomplete if people cannot discover what happened to their concerns. In the fictional exercise, the committee’s reply should identify the chosen arrangement, the reasons it found persuasive, the objection it has not resolved and the next review point. It should distinguish a decision made now from a possibility still under investigation.

Students can draft two sentences responding directly to Dev: “We understand that Thursday is your only available project time. The proposed cycle would retain two Thursdays, so it would reduce your opportunities; we need to establish whether that loss can be addressed before changing the calendar.” This acknowledges the consequence without claiming that a short explanation compensates for it.

For the trial record, keep attendance for every slot, requests for places that cannot be accommodated, and short accounts of difficulties created by the change. Agree what those records mean before interpreting them. Ten requests from one person for different dates are different from ten people requesting the same session. The question determines which distinction matters.

Review more than the headline total. A rise in attendance could coincide with loss of access for someone who relied on Thursday. A stable total could conceal a useful change in who can participate. The government and public intelligence article develops the institutional responsibility to explain and revisit decisions. Here, students practise a small, explicit version of that responsibility.

Contents · Previous chapter · Next chapter

Collect a short first response before group discussion, one revision after examining the records, and a later individual response to changed conditions. These show different stages of the learner’s work. A polished group recommendation cannot by itself identify who understood the attendance unit, noticed Dev’s constraint or proposed the review question.

The collaborative learning article explains why shared work needs individual evidence. Here, a brief oral explanation, annotated table or written paragraph can reveal reasoning that a confident spokesperson might otherwise conceal. Record substantive hints separately from help with reading, access or expression.

Contents · Previous chapter · Next chapter

Give each learner this new fictional operations note: the Wednesday room slot will be unavailable throughout the next four-week cycle. No replacement slot is available. The other five slots, their capacity and their duration remain unchanged. Thursday remains the only activity that the panel may change. The earlier register still describes the previous four weeks; do not alter those historical observations to fit the new timetable.

Ask the learner to reconsider A, B and C under the new constraint. They should calculate the quiet and project provision over four weeks, check the minimum of two weekly sessions for each activity, and explain how the closure changes the argument. Then ask for a short recommendation that answers one serious objection and identifies one unanswered question.

Keep the original records available. The target is reasoning from evidence, not remembering a table. A calculator, accessible text or assistance recording an answer can remain where appropriate to the learner. Withhold the earlier model recommendation and any new AI-generated conclusion during the first attempt. If a teacher supplies a missing session count or a reason to favour an option, record that help.

Do not demand a changed final preference simply because the facts changed. A learner may retain a position after examining the new trade-off. What matters is whether the explanation now addresses the unavailable Wednesday slot, rather than repeating the earlier justification unchanged. The learning transfer article explains the value and limits of such a nearby adaptation task.

Contents · Previous chapter · Next chapter

The closure removes four quiet sessions from every option. A now supplies eight quiet sessions and twelve project sessions: 160 and 240 place-visits respectively. B supplies twelve quiet and eight project sessions: 240 and 160 place-visits. C supplies ten sessions for each activity: 200 place-visits for each. All three have twenty sessions, forty room hours and 400 place-visits across the four weeks.

Each option still meets the panel’s minimum. Under A, every week has two quiet and three project sessions. Under B, every week has three quiet and two project sessions. Under C, Weeks 1 and 3 follow A’s pattern; Weeks 2 and 4 follow B’s pattern. This weekly check is necessary: a satisfactory four-week total alone would not prove that a weekly condition had been met.

Historical quiet use was 216 attendances. A’s new quiet capacity of 160 cannot accommodate that many quiet attendances over the cycle, even if every remaining place is filled. The difference is 56 potential attendance opportunities under that comparison, not 56 identified people who will be excluded. C’s quiet capacity is sixteen below the historical total. B’s exceeds it, but availability at the remaining times is still unknown.

The closure therefore strengthens a reason to consider more Thursday quiet provision. It does not erase Dev’s constraint or establish that B is the only defensible choice. A learner could support B while acknowledging the Thursday project loss and requesting an alternative opportunity; another could retain C as a limited trial while explaining the tighter quiet capacity. Both need a reasoned account of the consequences they accept.

Contents · Previous chapter · Next chapter

A teacher can run the enquiry in stages. First, give students the question and register, and ask for one observation and one uncertainty. Then introduce the comments and operations note. Let students revise the question they would ask the committee before presenting the AI-generated alternative. This makes a change in their thinking visible without requiring a lengthy record of every conversational turn.

Assign rotating jobs such as checking a source, explaining an affected person’s stated condition and identifying the strongest objection. Students should eventually practise more than the role that feels easiest. A learner who adds numbers accurately may need practice explaining a value; a fluent speaker may need to trace a factual statement back to the register.

Provide the complete pack in a readable, accessible form. A student can listen to a record, dictate a response or use a structured note sheet while still making the central judgement. A sentence starter such as “I favour this because…” supports expression; a prefilled reason claiming that C is fair supplies part of the thinking. Record the distinction when interpreting the work.

Parents can ask which condition matters most to the child and what evidence could change the recommendation. They need not disclose family political views or turn the activity into a contest over the right opinion. A calm request to explain one table entry or fairly restate an objection is enough to reveal a productive next learning step.

Contents · Previous chapter · Next chapter

End with a concise decision record: the question considered, the evidence used, the priority applied, the proposal supported, the objection retained and the next review condition. This is an original classroom device, not a certificate of general civic competence. Its value is that another learner can follow the reasoning and identify a point that needs correction or further evidence.

A later discussion might concern a shared equipment timetable or the use of a school display area. Begin by finding the actual constraints and affected people again. Do not transfer the room’s twenty-place capacity, alternating-week answer or minimum-session rule into a setting where they have no basis. The transferable habit is examining relationships between evidence, priorities and consequences.

AI can contribute at several points: proposing a feasible alternative, organising a comparison, identifying an unanswered question or helping a learner express a reason more clearly. Those contributions become educationally useful in this proposed sequence when the learner examines them, makes consequential choices and can explain the result. A fluent recommendation alone cannot establish that learning occurred.

The larger civic purpose is participation that improves shared understanding. Students should be able to say what they know, what they value, whose stated concern remains unanswered and why they are willing to revise a proposal. Agreement may emerge, but a clear record of an unresolved disagreement can also help the next decision. A community gains something when its members can contribute reasons that others can inspect and answer.

Contents · Previous chapter · Continue the series