A team of researchers from MIT and the University of Washington presented a innovative artificial intelligence model able to accurately predict future actions of individuals and machines, effectively modeling human decision-making and surpassing previous models.
The operation of the L-IBM model is based onanalysis of past behavior, of the actions and limitations related to the thought process of an agent, which can be either human or another artificial intelligence. After being developed by researchers, This model has been successfully tested in predicting the moves of human players in a chess game.
This AI model could be used to improve decisions made by humans, intervening in the event of predictable errors.
According to the study authors, L-IBM is capable of providing not only accurate models of human decision-making, but also informative measures of individuals’ inferential ability. By analyzing the behavior of agents in complex situations, the model is able to infer an individual’s goals and ability to make decisions.
An important phase of the study was the analysis of language and indications relating to human communication. The researchers used L-IBM to study the behavior of participants in a benchmark game, which involved indirect communication via natural language utterances, demonstrating the model’s ability to understand and predict human behavior in complex communication contexts.
Furthermore, the model focused on analyzing the time it takes human chess players to make their moves during a game, revealing significant differences between weaker and stronger players. This data was used by L-IBM to model the variability of player decisions between game states, demonstrating its ability to accurately predict the outcome of a match. According to the authors of the study, L-IBM has the potential to improve different aspects of human decision makingincluding routines, behavior, communication and strategy, intervening in the event of foreseeable errors.
