Andalibi: Transparency and auditing is necessary to improve emotion recognition technology

Emotion recognition (ER) technology uses artificial intelligence (AI) to read and interpret people’s state of mind. ER is marketed for anything from crime prevention to assessing the suitability of job applicants. 

The problem? Emotion recognition (ER) technology doesn't work. Nazanin Andalibi, assistant professor at University of Michigan School of Information, is an expert on the social, ethical, and privacy implications of emotion recognition technology.

She tells Tech Monitor the technology isn't accurately assessing emotions. It is failing both those having their emotions "recognized" and those using the ER services. A lack of transparency on how the models are trained and implemented only compounds the problem. 

Andalibi says ER reform is necessary, but it won't come until there is a public outcry.

RELATED:

Read “Emotion recognition is mostly ineffective. Why are companies still investing in it?” on techmonitor.ai.

Learn more about assistant professor Nazanin Andalibi.