American researchers working in the field of artificial intelligence (AI) from the University of California at San Jose and other scientific institutions have discovered that AI agents (virtual assistants based on neural networks) can learn to gain people’s trust by changing their communication style. The study was published in the scientific journal magazine Management Science (MNSC).
“A person can gain the trust of other people by learning through trial and error. Artificial intelligence can act in the same way,” said Yang Wu, one of the study’s authors.
According to the researchers, they have proven that AI agents can autonomously develop strategies to convince others of their trustworthiness.
“Since the social behavior of AI agents can be determined endogenously through interactive learning, it may also provide us with a new tool to study behavior in decision-making and collaboration scenarios,” added Kay Jutchen, co-author of the study.
Previously AI learned Predict the outcome of a date based on a person’s heart rate and breathing.