Given the importance of understanding how AI works, how can we take core ideas of AI, such as computer perception, and how computers “sense” their environments, and make them visible for students of all ages? Teaching AI literacy with a robot is an ideal way to make AI concepts tangible for students. Students can use sensors on a robot to collect data and use this data to solve problems and make decisions in real world contexts.

For example, young students can begin learning about computer perception—the extraction of meaning from sensory information—by exploring how sensors on robots interpret the world around them. When they discover that sensors like the Eye Sensor on VEX 123 or VEX GO read color as a numerical hue value, rather than the nuanced way humans perceive light, they can directly compare human and robot perception. This builds an early understanding that computers “see” differently from people. As students investigate further and notice how changes in lighting affect those hue value readings, their understanding deepens. They begin to recognize not only how sensor perception works, but also its limitations and the conditions that influence AI-driven interpretations of the world.

As students move into upper elementary and middle school, they can extend their understanding of sensor data through AI Vision on platforms like VEX AIM, IQ, EXP, and VEXcode VR. Hands-on experiences with AI Vision help learners begin to see that computer perception involves classifying and interpreting sensor data. This opens the door to deeper conversations about accuracy, reliability, and the limits of AI.

For example, in the Trick the Sensor activity (available for VEX AIM in the VEX AIM Intro Course, as well as for VEX IQ and EXP in the 2025 Hour of AI Activities), students investigate how AI Vision identifies objects by deliberately trying to “trick” it. They test drawings, classroom items, and 2D or 3D shapes in different field locations to see whether the robot detects them and how it labels them. When students discover that the sensor confidently misidentifies certain objects or loses track of them depending on angle, size, or distance, they begin to ask exactly the kinds of questions that build AI literacy: Why did it think a marker was a barrel? What data is the AI relying on? How could we improve its ability to classify correctly? Students can follow up on this experience by watching this video about how the AIM AI Vision Sensor is trained. These investigations help students understand that classification is a data-driven process—not magic—and that humans play an essential role in shaping how AI systems interpret the world.

Students can also practice these skills in VEXcode VR, where they can explore AI Vision on the virtual Hero Bots in the V5RC and VIQRC Playgrounds. Each virtual Hero Bot is pre-trained to detect game objects and report information such as height, width, and x/y coordinates. Students must analyze this data and transform it into strategic decisions—determining how to navigate, when to pick up an object, or how to score the most points in a time-based challenge.

In V5RC Push Back and VIQRC Mix & Match, students learn that AI doesn’t give a robot instant or perfect awareness. The robot must gather sensor data, interpret it, and act on it through the logic students create. Errors in detection or interpretation invite students to refine their code, building the understanding that AI systems must be tested, validated, and improved over time.