Scientists at the Massachusetts Institute of Technology (MIT) have developed a new method for combatting belief in conspiracy theories using artificial intelligence. The experiment involved 2190 participants who acknowledged their belief in various conspiracy theories, and the results revealed a 20% decrease in belief in the theories.
During the experiment, participants were asked to provide a brief description of the conspiracy theory they believed in and then rate their confidence in it. They then engaged in a conversation with artificial intelligence (GPT-4 Turbo), with each side presenting 3 arguments. Following the discussion, participants were asked to reassess their confidence in the truth of the theory.
The AI model took a direct and fact-based approach, using rational arguments in favor of alternative points of view in 83% of conversations. Unlike humans, the AI avoided emotional reactions that could trigger defensive responses and instead focused on neutral and patient communication.
As a result of the experiment, belief in a specific conspiracy theory decreased by approximately 20%, and there was a general decrease in belief in other conspiracy theories. The impact of the AI interaction lasted for at least 2 months, suggesting a lasting change in participants’ beliefs.
MIT researchers highlighted that previous attempts to debunk conspiracy theories failed due to a generalized approach to facts. The AI’s deep and precise analysis was found to be more effective in influencing participants’ consciousness.
John Sinnotta from the University of Huddersfield emphasized the importance of the study, stating that this approach could significantly impact the views of individuals inclined to believe in conspiracy theories.