Short answers, pulled from the story.
Rosalind Picard founded modern affective computing with her 1995 paper "Affective Computing" and her 1997 book of the same name, published by MIT Press. The field sits at the intersection of computer science, psychology, and cognitive science.
Speech emotion recognition systems achieved an average reported accuracy of 70-80 percent in research from 2003 and 2006. This exceeds average human accuracy, which is approximately 60 percent, though systems that combine speech with physiological data or facial expressions perform better.
Paul Ekman proposed six basic emotions in 1972: anger, disgust, fear, happiness, sadness, and surprise. He later expanded this list in the 1990s to include emotions such as amusement, contempt, contentment, embarrassment, excitement, guilt, pride in achievement, relief, satisfaction, sensory pleasure, and shame.
The Facial Action Coding System (FACS) was created by Paul Ekman and Wallace V. Friesen in 1978, based on earlier work by Carl-Herman Hjortsjö. It maps specific combinations of facial muscle contractions, called action units, onto emotional expressions. For example, happiness corresponds to action units 6 and 12, while fear requires units 1, 2, 4, 5, 20, and 26.
Key risks include users developing parasocial relationships with AI systems, which can deepen depression or loneliness in people with mental health conditions. Additional concerns include analyzing facial expressions in public without consent, using systems to manipulate audience emotions, and potential psychological effects of affective AI in sexual contexts.
Rosalind Picard's cognitivist approach treats emotion as an objective internal signal that machines can recognize, express, and simulate. The interactional approach, associated with Kirsten Boehner and others, views emotion as culturally grounded and socially constructed, aiming to help people understand their own emotions rather than mapping feelings into mathematical models.