Artificial intelligence managed to convince conspiracy theorists of the falsity of conspiracy theories with the help of counterarguments and evidence. He is more adept at this than humans, because he can process large amounts of information and quickly provide rebuttals.In the experiment, it turned out that dialogues with an artificial intelligence model reduced people's belief in conspiracies by about 20 percent, and the effect persisted even two months after the end of the study. Such results were published in the journal Science.

How was the effectiveness of AI tested?

The researchers used the GPT-4 Turbo artificial intelligence model and recruited 2,190 participants who believed in conspiracy theories, including the assassination of US President Kennedy, the origin of covid and evidence of aliens on Earth. Research participants described the theories they believed in, and the AI ​​was able to provide reasoned counter-arguments that were 99 percent true by a professional fact-checker.

As a result of communication with the AI, 27.4 percent of the participants became unsure of their beliefs, and partly even in those theories that were not discussed with it. The effect was noticeable among participants with different kinds of conspiracy beliefs and even among those who were convinced that they were right at the beginning of the experiment. This study challenges the common belief that belief in conspiracy theories is difficult to change with facts and arguments.

• Previously, scientists had already developed a game that was supposed to counter conspiracy theories. Interestingly, it was also useful for teaching people to recognize the manipulations of ISIS terrorists, which they use to try to convert people to their ranks.
• We talked in more detail about why people believe in conspiracy theories and various kinds of manipulation in our article "Theory of Lies".

Source