Conversations with AI chatbots can reduce and reverse conspiracy theory beliefs

The saying goes, “Once a person is down the rabbit hole of conspiracy theories, they’re lost for good,” describes the typical consensus regarding conspiracy theorists. Well, that idea is being overturned by artificial intelligence (AI).

An intriguing research study has demonstrated that even the staunchest conspiracy theorists can shift their views following brief interactions with AI chatbots.

This particularly holds true for those who hold deep-seated beliefs in some of the most firmly held conspiracies, such as the 2020 U.S. presidential election fraud and the COVID-19 pandemic.

Fueled by political polarization, social media, and widespread misinformation, conspiracy theories have grown into a significant public concern. They have managed to drive a wedge between theorists and their close friends and relatives.

A YouGov survey conducted last December showed a large number of Americans succumbing to various baseless conspiracies.

“Surprising but convincing” evidence

The study presents a strong challenge to the accepted view within psychology that conspiracy theorists stick to their beliefs because they hold value to their identities or resonate with their underlying motivations.

Thomas Costello, assistant professor of psychology at American University and the lead author of this study, is of the view that most previous approaches have focused on preventing people from forming these beliefs, rather than rectifying them.

“Many conspiracy believers were indeed willing to update their views when presented with compelling counterevidence. I was quite surprised at first, but reading through the conversations turned me into a believer,” Costello enthused.

“The AI managed to provide lengthy, highly detailed reasons debunking the claimed conspiracy in each round of conversation. It also proved adept at establishing cordial rapport with the participants.”

AI chatbots and conspiracy theories

The study comes at a time when society is in the middle of a debate on the potential benefits and risks of AI and chatbots.

Large language AI models serve as rich repositories of knowledge, and the authors of this research emphasize one way in particular that these vast databases can help people form more accurate beliefs.

Artificial intelligence used in chatbots can gather data from diverse topics within seconds, making it possible to tailor counterarguments against specific conspiracy theories in ways that humans simply can’t match.

Gordon Pennycook, associate professor of psychology at Cornell University and co-author of the paper, echoes this sentiment.

“Previous attempts to debunk dubious beliefs have a major limitation: One has to guess what people’s actual beliefs are in order to debunk them — not an easy task,” Pennycook remarked.

“In contrast, the AI is designed to respond directly to people’s specific arguments using strong counterevidence. This offers a unique chance to assess just how responsive people can be to counterevidence.”

Chatbots are designed to be as persuasive as possible while engaging participants in tailored dialogues. GPT-4, the AI model powering ChatGPT, provided factual responses to participants’ conspiratorial claims, which gives the models more credibility.

Results and implications

The study’s results are quite impressive, to say the least, offering a sense of optimism in the realm of psychology and cognitive science.

Participants exhibited a notable shift in their beliefs after engaging with the AI, demonstrating a reduction in conviction towards their previously held conspiracy theories.

This shift suggests that even the most entrenched beliefs may not be as immutable as once thought.

The implications extend beyond academic circles. They prompt a re-evaluation of strategies used in public health campaigns, educational efforts, and policy-making aimed at combating misinformation.

By harnessing AI’s potential to engage individuals with tailored, evidence-based dialogues, society can make strides toward promoting more informed and rational public discourse.

Next steps for researchers

The researchers acknowledge that the study is just the beginning, and many avenues for future research remain.

One area to explore is the long-term effects of AI-facilitated belief modification. While the study showed promising short-term results, understanding whether these changes persist over time is crucial.

Additionally, future studies could investigate the effectiveness of AI in addressing conspiracy theories across different cultural and linguistic contexts, as well as its potential adaptability to various mediums, such as social media platforms or educational tools.

Expanding the scope of AI’s application may reveal broader opportunities for its integration into societal efforts aimed at fostering critical thinking and evidence-based reasoning.

Conspiracy theories, AI, and the future

To sum it all up, these results presented by Thomas Costello and his team are encouraging and suggest a future where AI could play a role in helping reduce conspiracy beliefs when used responsibly.

Nonetheless, further studies will be required to determine long-term effects, using different AI models, and practical applications outside of a lab setting.

David Rand, a paper co-author and professor at the MIT Sloan School of Management, is optimistic about this potential.

“Although much ink has been spilled over the potential for generative AI to supercharge disinformation, our study shows that it can also be part of the solution,” said Rand. “Large language models like GPT4 have the potential to counter conspiracies on a massive scale.”

—–

For those interested in engaging with this ongoing work, there is a website available where the public can try out the intervention for themselves.

The study is published in the journal Science.

—–

Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.

—–