Agents of ViTAL: Ethics Missions -- A Narrative-Centered Learning Environment with a Co-Designed Conversational Agent for Middle School AI Ethics

In the modern classroom, technology is no longer just a tool for delivering lessons; it is becoming a participant in the learning process itself. Artificial intelligence systems are increasingly capable of analyzing student behavior, offering feedback, and even predicting future performance. While these tools promise efficiency and personalization, they also introduce complex questions about privacy, fairness, and the hidden costs of computing power. For young people growing up in a world saturated with algorithms, understanding how these systems work and what ethical trade-offs they involve is becoming as essential as reading or writing. The challenge for educators is not simply to teach the mechanics of code, but to help students navigate the difficult decisions that arise when technology meets human values. This requires a space where learners can step into the shoes of decision-makers, weigh conflicting interests, and practice the art of ethical reasoning in a setting that feels real, even if the scenario is imagined.

A team of researchers from universities across the United States has developed a new way to bring these abstract concepts into the middle school classroom. They created a digital learning environment called "Agents of ViTAL: Ethics Missions," a browser-based experience where students work together to solve a fictional problem. In this scenario, a middle school is considering whether to adopt a new artificial intelligence system called PresenterPal, designed to provide instant feedback on classroom discussions. The students are not passive listeners; they are the decision-makers. They enter a virtual space that blends the quiet order of a library with the experimental nature of a laboratory, known as the "Libratory." Here, they meet digital guides who set the stage for their mission: to investigate whether the school should move forward with this technology.

Once the investigation begins, the students explore a simulated version of the school, gathering information from various people who would be affected by the new system. They speak with teachers, students, administrators, and IT staff, collecting evidence such as news reports about data breaches or notes on budget constraints. As they move through the simulation, they encounter specific situations that force them to confront the three main ethical dimensions of artificial intelligence: privacy, bias, and environmental impact. For instance, when configuring the system, students must decide whether to store student interaction data on local servers within the school or on cloud servers elsewhere. This choice is not merely technical; it carries weight. Storing data locally might protect privacy but strain the school's limited budget, while using the cloud might offer better security but raise concerns about who else can access the information. These moments are designed to show that there is rarely a perfect solution, only a series of difficult compromises.

After gathering their evidence, the students move to a collaborative phase known as the Ranking Challenge. In this part of the experience, the group must come to a consensus on which risks and benefits of the AI system are most important. They use a shared digital board to rank these factors, discussing their reasoning in an integrated chat window. This board acts as a visual record of their collective thinking, helping them see how their individual views shift as they listen to one another. The goal is not to find a single "correct" answer, but to practice the skill of justifying trade-offs using the evidence they found earlier. They must articulate why they value privacy over convenience, or why they believe environmental costs matter more than immediate efficiency, and then agree on a final recommendation for the school.

Supporting this entire process is a digital companion called EthicsBot. Unlike a traditional teacher or a search engine that provides answers, EthicsBot is designed to act as a peer-like thought partner. It does not tell the students what to think; instead, it asks questions that prompt them to consider alternative perspectives, clarify their own reasoning, and reflect on the ethical weight of their choices. The system was built with a unique approach to its creation: the students themselves helped design the bot. Through a process of co-design, high school students provided feedback that shaped how the bot speaks, what it suggests, and the safety limits it follows to prevent it from giving misleading information. This ensures that the tool feels relevant to the learners and respects their voice in the development of the technology they are studying.

The researchers tested this environment with four ninth-grade classes, involving eighty students who had given their consent to participate. The results showed that the students were deeply engaged with the material. They did not just play a game; they demonstrated a capacity for substantive ethical reasoning, connecting the fictional scenarios in the simulation to their own real-life experiences with technology. The students were able to discuss complex issues like data security and algorithmic bias with a level of nuance that suggests the narrative approach is effective. By placing students in the role of active participants rather than passive recipients of information, the system helps them build the confidence and skills needed to navigate an increasingly automated world. The project suggests that when young people are given the space to explore these dilemmas collaboratively, they can develop a sophisticated understanding of the ethical landscape of artificial intelligence.