The program, called Class, is AI-based and works with Zoom to analyze students’ faces in real time. It’s said to be capable of detecting three emotional states: bored, confused, and distracted. Intel and Classroom Technologies state the program will “give [teachers] additional insights to allow them to better communicate,” thus solving the modern problem of disengagement in virtual classroom environments.
Class works by capturing images of students’ faces using computer vision technology, then comparing those images against psychological data and current context. Despite the fact that people tend to express their emotions in different ways (especially across cultures and abilities) Intel denies that the resulting labels could carry “any assumptions,” as the company only trained the AI to see an expression a certain way if two separate psychologists agreed on the emotion being displayed.
This by nature will require students to have their cameras turned on, which (as others have pointed out) has both financial and social implications. Keeping one’s camera on during a video call uses up more bandwidth, which can be a problem for those with a mobile hotspot or limited monthly usage. Revealing one’s physical background to others can also expose one’s socioeconomic or family situation, which some find uncomfortable and unnecessarily intimate.
What’s unclear is what will happen once Class identifies a student as bored, disengaged, or confused. Will the instructor message them privately? Will the student be required to undergo additional classwork until they’ve been deemed no longer confused? Though this is likely up to the instructor’s discretion, it’s worth asking whether students will be inconvenienced—or worse, made to feel uncomfortable or unworthy—due to microexpressions they can’t really control. That’s if instructors even want to use the technology, of course; some told Protocol they’d be against implementing Class and would be wary of any employer that required it. “I think most teachers, especially at the university level, would find this technology morally reprehensible,” one said.
Class’ expression-analyzing technology is still in the testing phase. In the meantime, the software currently offered by Classroom Technologies already auto-verifies students’ identities and ranks students by how much they talk in class.