A debate game about societal impacts of Artificial Intelligence

To address the public's limited understanding of AI, the authors propose and validate a free, interactive debate game that simulates a municipal council selecting AI solutions, effectively raising awareness about AI's societal impacts while fostering students' argumentation and listening skills.

The Gist

Imagine you are walking into a high school classroom, but instead of a standard math or history lesson, the room is buzzing with the energy of a town hall meeting. The students aren't just sitting in rows; they are sitting in groups, wearing invisible "hats" that change who they are for the next hour.

This paper describes a game designed to teach teenagers (and adults) about Artificial Intelligence (AI) not by teaching them complex code, but by making them argue, debate, and vote on how AI should be used in their lives.

Here is the breakdown of the game, the results, and why it matters, explained simply.

🎭 The Game: "The AI Town Council"

The Setup:
Imagine a fictional town called "Wafer" is facing a scary epidemic (like the flu or COVID-19). A tech company called "Sowana" comes to the town council with three different AI solutions to stop the virus. The students are split into groups, and each group is assigned a specific character role they must play, regardless of their own real-life opinions.

The Characters (The "Hats"):

• The Centrals: Busy city workers with money but no time. They want things to move fast and keep their freedom to travel.
• The Alters: Young, eco-friendly activists. They are skeptical of new tech and want guarantees before they trust it.
• The Seniors: Retired people who just want to stay healthy and independent.
• The Futurists: Tech-loving, online-savvy kids who love innovation but have no money.
• The Remote: People who live far away and have to commute; they just want to get to the city easily.

The Three AI Solutions (The "Menu"):

1. Eye'Wana: A giant army of drones and cameras with facial recognition to catch rule-breakers and punish them automatically. (The "Big Brother" option).
2. Wana'Like: A social media app that suggests safe places to go and gives you coupons for discounts. (The "Friendly Guide" option).
3. Wana'Pass: A health passport app that tracks your body temperature and medical data, giving you points to decide how much you can go out. (The "Health Tracker" option).

The Mission:
The groups must debate which solution is best for their specific character. They have to weigh the pros and cons: Is it effective? Does it invade privacy? Is it fair? After arguing, they vote. Then, they have a second round of debate where they can try to change each other's minds before a final vote.

🧠 Why Play a Game Instead of a Lecture?

The authors (Carole and Cédric) realized that most people think AI is either a magic robot that saves the world or a scary monster that destroys it. They wanted to show that AI is actually just a tool, and like any tool, it has a price tag and a side effect.

• The "Role-Play" Trick: By forcing a student who loves privacy to play the "Futurist" who loves surveillance, the game forces them to step out of their own skin. It's like acting in a play; you learn more about the character's fears and hopes when you have to speak their lines.
• The "Debrief" (The After-Party): The game doesn't end with the vote. The teacher leads a discussion afterwards to connect the game to the real world. They explain, "Hey, that 'Eye'Wana' game? It's actually happening in some countries right now." This turns the game into a mirror for reality.

📊 What Happened When They Played?

The authors tested this game with 81 high school students and found some fascinating results:

1. It Wasn't About "Hard" Facts: Students didn't leave knowing how to code a neural network. Instead, they left with "Soft Skills." They learned how to listen to someone they disagree with, how to build an argument, and how to realize that there is rarely one "perfect" answer to a societal problem.
2. The "Frustration" Factor: Some students got frustrated because there was no "right" answer. Some solutions were all a bit "bad." The authors say this is actually a good thing. It teaches citizens that real life is messy, and we often have to choose the "least bad" option while trying to protect our rights.
3. The Surprise Winner: When students voted based on their own personal opinions, they mostly liked the friendly app (Wana'Like). But when they voted based on their assigned roles in the game, the town council often picked the scary camera system (Eye'Wana). This proved that the students could successfully "act" and defend a viewpoint that wasn't their own.
4. The Fear Factor: After playing, many students expressed a healthy amount of fear or suspicion about AI. They realized, "Wait, this tech could be used to spy on me." The goal wasn't to make them hate AI, but to make them wake up and ask questions.

🏁 The Big Takeaway

Think of this game as a flight simulator for citizenship.

Before you get your pilot's license, you don't just read a book about aerodynamics; you sit in a simulator and practice handling turbulence. Similarly, before we let AI make big decisions about our health, jobs, and privacy, we need to practice making those decisions in a safe, low-stakes environment.

This game shows that you don't need to be a computer scientist to understand AI. You just need to be a curious human who is willing to argue, listen, and ask: "Who benefits from this, and who pays the price?"

The game is free to download, and the authors hope that by playing it, we can all become smarter, more critical citizens in a world run by algorithms.

Technical Summary

1. Problem Statement

Artificial Intelligence (AI) is increasingly ubiquitous in decision-making processes (e.g., CV screening, autonomous driving, public security), yet the general public lacks a fundamental understanding of how these algorithms function, their reliance on personal data, their inherent biases, and their fallibility.

• The Gap: While AI is pervasive, public literacy regarding its mechanisms and ethical implications is low. The general population is often swayed by media narratives portraying AI as either infallible or catastrophic, without the ability to weigh benefits against risks.
• The Need: In alignment with UNESCO recommendations, there is an urgent need to educate non-specialists to enable informed choices regarding data delegation and technology adoption. Traditional educational methods (e.g., MOOCs) often leave learners passive and fail to engage the general public or high school students who lack technical prerequisites.

2. Methodology

The authors developed and evaluated a serious game (a debate simulation) designed to foster critical thinking and argumentation skills regarding AI's societal impacts.

Game Design & Mechanics

• Scenario: The game simulates a municipal council meeting convened to address a societal crisis (specifically, an epidemic). A fictional company, "Sowana," proposes three distinct AI-based solutions:
  1. Eye'Wana: A surveillance fleet (cameras/drones) using facial recognition to detect and punish rule violations.
  2. Wana'Like: A recommendation app linked to social media profiles, suggesting uncrowded, safe locations in exchange for discounts.
  3. Wana'Pass: A points-based health passport connected to medical records and physiological sensors, restricting movement based on health status.
• Roles: Participants are randomly assigned one of five stereotyped personas representing diverse societal viewpoints:
  • Centrals: Downtown workers valuing mobility and time.
  • Alters: Eco-conscious/anti-globalization youth.
  • Seniors: Retirees focused on autonomy.
  • Futurists: Tech-savvy youth with limited income.
  • Remotes: Heterogeneous group living outside the city center.
• Process: The 90-minute session follows eight phases:
  1. Random role assignment.
  2. Presentation of the agenda and solutions.
  3. Group discussion (pros/cons based on the assigned role).
  4. Initial scoring (distributing 5 points among solutions).
  5. Tabulation of scores and arguments.
  6. Open debate to shift perspectives.
  7. Re-scoring based on the debate.
  8. Debriefing: Crucial phase linking game concepts to real-world technologies, legal frameworks (GDPR), and ethical implications.
• Evaluation Criteria: Participants evaluate solutions based on a grid inspired by privacy experts, covering: Effectiveness, Impact on Individual Liberties, Impact on Privacy, Economic/Ecological Cost, Accessibility/Discrimination, and Risk of Errors.

Evaluation Methodology

• Data Collection: The game was tested in five sessions with high school students (N=81). Data was collected via paper questionnaires distributed immediately after the sessions to capture "hot" feedback.
• Analysis: Quantitative analysis (Likert scales) and qualitative analysis (word frequency in open-ended responses) were performed using Python scripts.
• Peer Review: The game was also presented to approximately 15 teachers and researchers for qualitative feedback on pedagogical biases and feasibility.

3. Key Contributions

• Pedagogical Tool: A freely available, low-barrier-to-entry debate game that requires no prior technical knowledge of AI, making it accessible to high schoolers and non-technical teachers.
• Role-Playing Framework: A mechanism to detach students from their personal biases by forcing them to argue from a specific, often opposing, perspective. This promotes empathy and the understanding that diverse, reasonable arguments exist for complex issues.
• Critical Thinking Focus: Unlike technical courses, this tool focuses on "soft skills": argumentation, active listening, weighing benefit-risk trade-offs, and exercising critical judgment on technology adoption.
• Addressing "Technosolutionism": The game forces participants to choose between imperfect solutions, highlighting that there is rarely a "perfect" answer to societal problems, thereby fostering a more nuanced view of technology.

4. Results

The evaluation of 81 participants yielded the following insights:

• Engagement & Perception:
  • The game was rated highly interesting (8/10 average).
  • 77 out of 81 participants reported feeling engaged.
  • The debate format and role-playing were the most appreciated aspects.
• Role-Playing Effectiveness:
  • While self-identification with roles was moderate (5/10 average), the ability to perform the role was high (8/10 average).
  • Notably, students successfully argued for solutions they personally disliked. In the final vote, the group chose Eye'Wana (surveillance) in 3 out of 5 sessions, whereas the students' personal preference was Wana'Like (37 votes vs. 18 for Eye'Wana). This confirms the game successfully separated personal opinion from role-based argumentation.
• Learning Outcomes:
  • Hard Skills: Low perceived acquisition of technical knowledge (6/10), which was expected as the game is not a technical course.
  • Soft Skills: High perceived impact on reflection (7/10) and critical thinking.
  • Behavioral Intent: 18/40 participants expressed a desire to research AI further, and 20/40 wanted to continue debating.
• Qualitative Feedback:
  • Positives: Students valued the freedom to debate and the ability to express opinions without fear of personal judgment.
  • Negatives: Some frustration arose from the lack of a "right answer" and the feeling that all solutions were flawed. Some students requested more technical details, which the authors clarify is outside the game's scope.
• Bias Analysis:
  • Critics noted the roles were somewhat stereotypical, but the authors argue this "cooling" of political identities helps facilitate debate.
  • The "technosolutionist" constraint (forcing a choice) was defended as a realistic simulation of democratic decision-making where imperfect choices must be made.

5. Significance

• Democratizing AI Literacy: The paper demonstrates that complex societal issues surrounding AI can be effectively taught to non-experts through gamification and role-play, bypassing the need for mathematical or coding prerequisites.
• Civic Education: It provides a model for training future citizens to navigate the "benefit-risk" balance of emerging technologies, a skill essential for democratic participation in an AI-driven society.
• Scalability: The game is designed for easy implementation by teachers without AI expertise, supported by a comprehensive facilitator guide and free downloadable materials.
• Future Directions: The authors suggest the framework could be adapted for other disciplines and themes, emphasizing that the core value lies in the process of critical debate rather than the specific topic of AI.

In conclusion, the paper validates the use of serious debate games as an effective pedagogical strategy to move beyond technical literacy toward societal and ethical literacy regarding Artificial Intelligence.