ChatGPT Helps Diminish Conspiracy Beliefs, Study Finds
A recent study has revealed that engaging in a brief conversation with ChatGPT, a chatbot powered by GPT-4, can significantly reduce belief in conspiracy theories. Researchers from institutions including the Massachusetts Institute of Technology found that conversations with the AI led participants to lower their belief in deeply entrenched conspiracies, such as those related to COVID-19, by about 20 percent on average.
The study, published in the journal Science, involved over 2,000 individuals who identified as conspiracy believers. These participants engaged in dialogues with ChatGPT designed to counteract their beliefs with factual information. The AI used its advanced capabilities to generate personalized responses and refute specific evidence cited by the participants.
The interaction, which lasted an average of 8.4 minutes and included three rounds of discussion, had a noticeable impact. Approximately one-quarter of the participants abandoned their previously held conspiracy theories after chatting with the AI. This reduction in belief persisted for at least two months following the conversation.
Thomas Costello, the lead author and assistant professor of psychology at American University, noted that many participants were willing to reconsider their views when presented with well-crafted counter-evidence. The AI’s ability to provide detailed, persuasive arguments and build rapport played a key role in this process.
The researchers highlighted that traditional methods of addressing conspiracy beliefs have struggled due to the vast range of theories and the high level of detail conspiracy theorists often possess. Unlike human debaters, ChatGPT can quickly tailor its responses to address specific arguments and counter-evidence effectively.
Co-author Gordon Pennycook, an associate professor of psychology at Cornell University, emphasized the advantage of AI in this context. He explained that ChatGPT’s capacity to provide targeted rebuttals offers a unique opportunity to challenge and potentially change conspiracy beliefs in ways that human efforts have yet to achieve.
As discussions continue about the role of AI in society, this study underscores the potential of advanced language models to address misinformation and foster critical thinking.
