When you were in middle school, what games did you play to help you learn? Maybe you played chess to develop strategic thinking or Scrabble to grow your vocabulary. If computers were around, maybe you played Oregon Trail to learn the history of the United States frontier or Where in the World is Carmen Sandiego? to test your knowledge of world geography.

A new free, gamified learning experience designed by the Stanford Accelerator for Learning and Google Research seeks to teach teens about the technologies of today, including artificial intelligence. The game, called AI Quests, transforms learning about how AI works into a problem-solving journey, where students collect data, build models, and understand how decisions they make influence the AI models.

“The research shows that even though we hear about ‘data’ all the time, middle school students, and frankly, a lot of older students, are not very familiar with what data are or how exactly they are selected and used in the process of developing or training an AI,” said Associate Professor Victor Lee, faculty lead for AI+Education at the Accelerator and one of the creators of AI Quests.

Associate Professor Victor Lee, faculty lead for AI+Education at the Stanford Accelerator for Learning, and a co-creator of AI Quests. | Google Research

Teens are increasingly using generative AI for schoolwork, but without an understanding of how it works and the role humans play in its design. Additionally, they may not know how to be critical consumers of AI or how it can be used for good.

“I hope AI Quests not only teaches [young people] new skills, but also sparks their creativity to dream up new ways to solve problems with AI,” said Yossi Matias, vice president and GM of Google Research.

“The AI technology you’ll experience is amazing, but it’s not magic,” says Professor Skye, the fictional character who guides students through AI Quests, in the video introduction to the game. “The universe is counting on you. Are you ready?”

Beyond the tech: what middle schoolers need to know

The teenage years are a time of identity development as children transition to adulthood. It’s also a time where young people increasingly connect their learning in the classroom to the world around them and empathize with people with different life experiences than their own.

In AI Quests, students take on the role of a researcher applying AI to a real-world scenario. Initial modules include flood forecasting and blindness prevention. With help from characters like Professor Skye and Dr. Visus, a subject matter expert, they collect, select, and clean data, and then use it to train and test their own AI model.

The game was crafted based on decades of research on learning design, incorporating elements of active learning, story-based learning, pedagogical agents (in-game characters), and in-game mini-assessments to check for understanding.

With help from in-game characters, students collect, select, and clean data, and then use it to train and test their own AI model. | Google Research

“In designing the Quests, we wanted to make sure students weren’t just being guided through how AI works, but were making choices and decisions and seeing the impacts of those decisions,” said Kristen Blair, director of research for the Accelerator’s Digital Learning Initiative and one of the game designers. “This creates a gameplay loop, and also a feedback loop, for students that really supports learning, where they grapple with a challenge, they think about how they want to solve it, they see the impact of that solution, and then they can revise their understanding.”

Beyond practicing empathy and working through complex real-world scenarios, the game helps students understand the role of human decision makers and subject matter experts in building and using AI models.

“Most often, teens are consumers of AI. In AI Quests, the roles are flipped, and they become creators or decision makers around AI,” said Blair. “They make decisions such as about what data to include or how to test their model, and this helps them to experience the role that humans play in the success of AI systems and the outcomes.”

“Even though the AI does tasks independently, in the end, it’s still a human that makes that decision,” said Ibrahim (‘Joba) Adisa, postdoctoral scholar at Stanford Graduate School of Education and the Stanford Institute for Human-Centered Artificial Intelligence (HAI), who also helped design the game. “And whether the AI does it perfectly or not … humans remain accountable.”

How teachers can use AI Quests

AI Quests were play tested with students, enabling the researchers to ensure an appropriate level of difficulty for the age group. The game was designed to be ready to use: it comes with lesson plans, supporting materials, and guides for teachers, ensuring seamless integration with existing curriculum.

“We designed it as a plug-and-play solution to empower educators to embed this learning experience however they see fit into the learning journeys they are leading,” said Ronit Levavi Morad, senior director and executive sponsor for AI literacy at Google Research.

“A lot of classroom teachers are interested in having conversations about AI with their students, but they don’t always feel prepared because this is such a new topic, and it’s moving very quickly,” said Lee. He and Adisa have been filling this gap through Classroom-Ready Resources About AI For Teaching (CRAFT), a database of lessons, including AI Quests, that support AI literacy.

“One of the unique things about AI Quests is it gives students an opportunity to see what is happening on the frontiers of AI research,” Lee said. “How do we solve complex human problems that have disparate kinds of data and many different complexities? Making sure that we address AI literacy now is really a down payment on the future of people’s lives, livelihoods, the economy, and their civic decision making.”