Intel AI student emotion monitoring system accused of being inaccurate, intrusive

Occurred: April 2022

An Intel AI-based software system that aims to detect the emotional states of students during digital classes sparked debates on AI, science, ethics, and privacy.

The solution, integrated with Zoom via the “Class” software product, analyses students’ body language and facial expressions to infer mental states such as boredom, distraction, or confusion and is intended to aid teachers in adapting their teaching techniques.

Some teachers who tested Intel's system in a physical classroom said the system was useful. However, critics of affective computing argued emotion recognition is notoriously inaccurate and often mistakes facial expresssions for real, underlying feelings. The technology is also known to struggle across different cultures and scenarios.

Others are concerned about the intrusive nature of using cameras in an educational context, and in their potential for continual surveillance and misuse.

Classroom Technologies co-founder and CEO Michael Chasen told Protocol he hopes to partner with one of the colleges his company works with to evaluate Intel's system.

Operator: 
Developer: Intel; Classroom Technologies
Country: USA
Sector: Education
Purpose: Improve student engagement
Technology: Emotion recognition; Facial analysis
Issue: Accuracy/reliability; Privacy; Surveillance
Transparency: Black box

Research, advocacy 🧮

Page info
Type: Incident
Published: April 2022