Intel Develops AI Tool to Analyse Students’ Emotions

Intel has developed an AI programme to analyse facial expressions and body language of school pupils and university students learning online, and to determine their emotional state. This is reported by Tom’s Hardware.

Engineers developed the feature in collaboration with Classroom Technologies and integrated it with Zoom. The software uses video streams from lessons, transmitted to the algorithm along with contextual information in real time.

Intel researcher Sinem Aslan said that the main aim of the project is to improve individualised instruction, allowing the teacher to quickly respond to each student’s emotional state and steer the educational process in the necessary direction.

Many experts have expressed concerns about the use of such technologies in education. They say that modern algorithms cannot pick up the finest expressions of facial cues signaling changes in mental state. They added that emotions are expressed differently across cultures.

Intel notes that the model was built with the knowledge and experience of a group of psychologists who analysed video from real lessons recorded using laptops with 3D cameras. The experts then studied the data and annotated the detected emotions in order to form a training dataset for the algorithm.

In April 2021, researchers presented a game to show the risks of emotion recognition by artificial intelligence.

In the same month, the face-recognition algorithms used to monitor students during exams were accused of bias against Black students.

In August, a group of international researchers found that artificial intelligence is not capable of recognising human emotions.

Subscribe to ForkLog AI news on TikTok!