MIT study shows AI chatbots can reduce belief in conspiracy theories by 20%

The rise of conspiracy theories on the internet has become a major problem, with some theories leading to significant harm and misinformation. A recent study from MIT Sloan and Cornell University suggests that AI chatbots could be a powerful tool to combat these false beliefs. The study, published in the journal Science, shows that conversations with a large language model (LLM) such as GPT-4 Turbo can reduce belief in conspiracy theories by about 20%.

How AI Chatbots Work

The researchers, including Dr. Yunhao Zhang of the Institute for the Psychology of Technology and Thomas Costello of the MIT Sloan School of Management, tested the effectiveness of an AI chatbot by engaging 2,190 participants in text conversations about their favorite conspiracy theories. The AI was programmed to provide persuasive, fact-based rebuttals to each theory. The study showed that participants who interacted with the chatbot reported significantly less trust in the theories.

Accuracy and Future Impact

The study also ensured the accuracy of the chatbot’s responses by having professional fact-checkers review the claims made. Almost all (99.2%) of the claims were accurate, demonstrating the reliability of the information provided by the AI. The findings suggest that AI chatbots can be used across a variety of platforms to challenge misinformation and encourage users to think critically.

Next step

While the results are encouraging, further research is needed to explore the long-term effectiveness of chatbots in changing beliefs and addressing different types of misinformation. Researchers such as Dr. David G. Rand and Dr. Gordon Pennycook have highlighted the potential of integrating artificial intelligence into social media and other forums to enhance public education and counter harmful conspiracy theories.