Scientists from St. Petersburg State University (SPbU) together with specialists from the Artificial Intelligence Research Institute and the Institute of Psychology of the Russian Academy of Sciences, have tested how accurately modern language models can predict a person's psychological characteristics based on a standard interview. This was reported on the SPbU website.

The study involved 645 people. Each participant answered interview questions and completed psychological tests assessing 38 indicators, including personality traits, values, moral and social attitudes, as well as views on material possessions.

The scientists then provided the interview materials to five language models. The neural networks were tasked with predicting how a specific participant would answer the psychological test questions. The predictions were then compared with the actual responses of the participants.

The researchers found that AI is better at predicting psychological characteristics that are more clearly reflected in speech and self-description. For example, the models identified extraversion with relatively high accuracy.

Neural networks found it more difficult to determine levels of anxiety, the pursuit of pleasure and a person's views on fairness.

"In an applied context, such technologies can be used for personalized learning, psychological and motivational support, organizational communications and human interaction with intelligent systems. A separate promising area is modelling the behavior not only of individuals but also of entire teams," said SPbU Research Fellow Anastasia Panfilova.