AI

A woman's eye. / Photo by: Helenabdullah via Shutterstock

 

Defining the personality type of a person requires specialized tests, but what if it can be performed by AI software? In a joint effort, researchers were able to develop state-of-the-art machine learning algorithms that predict personality type based on eye movement.

The project was developed by experts at the University of South Australia, the University of Stuttgart, Flinders University, and the Max Planck Institute for Informatics which demonstrated how an advanced AI recognized the signs of the four of the Big Five personality traits – agreeableness, conscientiousness, extroversion, and neuroticism.

The AI software obtained data from a video eye tracker created by SensorMotoric Instruments. The AI tracked down the eye movements of participants in performing everyday duties in the university campus and analyzed these to define their personality type.

After that, the participants received questionnaires designed to assess personality traits. The questionnaires were used by researchers to cross-check the results from the AI.

“There’s certainly the potential for these findings to improve human-machine interactions. People are always looking for improved, personalized services. However, today’s robots and computers are not socially aware, so they cannot adapt to non-verbal cues,” said Dr. Tobias Loetscher, a researcher at UniSA.

Since the AI was able to get four out of the Big Five, the researchers validated the relationship between eye movements and personality traits. This means that certain eye movement patterns can be applied as predictors of specific personality traits.

The researchers noted that the findings also showed the difference between controlled studies in laboratories and real-life environments. The AI system was able to get more accurate information due to the natural responses of the participants during the study, more natural compared to lab settings.

The findings may be used to develop future computers and robots that possess near-human responses and social signals.