What if artificial intelligence could be used to counteract this trend? A recent study published in the prestigious journal Science explored this very possibility. A team of scientists used a Generative AI-based chatbotcapable of confronting people about their conspiracy beliefs and providing arguments based on facts and concrete evidence. The goal was to test whether AI could induce a change in the way people think, leading them to reconsider their beliefs.
The experiment came about because in the digital age, the spread of conspiracy theories and misinformation has become an increasingly pressing issue. The COVID-19 pandemic has further amplified this phenomenon, exploiting fear and uncertainty to spread false and misleading narratives.
The study of the research team, led by Thomas Costello of the American University of Washingtoninvolved over 1,000 participants, selected to represent a diverse sample of the US population. Each participant then interacted with the chatbot, based on the OpenAI’s GPT-4 Turbo language model and able to effectively address conspiracy theories, sharing a conspiracy theory they believed in and explaining the reasons behind their belief. The chatbot responded with detailed information and counter-arguments, highlighting evidence that disproved the conspiracy theory.
The results were surprising. Participants’ confidence in their conspiracy beliefs decreased by an average of 21% after interacting with the chatbot. Even more significantly, 25% of those who were initially strongly convinced of their theories showed signs of uncertainty after the AI interaction. This change in attitude persisted two months later, indicating a lasting impact of the chatbot interaction.
Despite the positive results, the study authors point out some limitations. The participants were paid users, and it is uncertain whether the same results can be replicated with people who have more deeply rooted and deep-seated beliefs in conspiracy theories. Additionally, while the chatbot has proven to be able to provide accurate answers, generative AI can still produce incorrect or misleading information. To ensure the reliability of the responses, the researchers submitted the chatbot’s responses to a professional fact-checker, who confirmed their accuracy and neutrality.
What do you think? Could the future of the fight against disinformation be marked by an alliance between human and artificial intelligence? Let us know in the comments below.
#fight #conspiracy #theories #chatbot #mind #hoaxes
