How AI can debunk a conspiracy theory in 8 minutes
Summary
Psychology professor Gordon Pennycook shares how he used an AI chatbot to debunk conspiracy theories in his brother, revealing that 25% of believers changed their minds after eight minutes of factual dialogue. The episode also explores "prebunking," a strategy by researcher Sander van der Linden that inoculates people against future manipulation by exposing common disinformation techniques. Together, these stories highlight that while truth can be persuasive, it requires unemotional, personalized evidence to overcome deep-seated beliefs and cognitive biases.
Topics discussed
Gordon Pennycook's brother falls down a conspiracy rabbit hole
Introduction to the show and the problem of changing minds
Why direct correction fails and the role of trusted influencers
The DebunkBot study: AI chatbot changes 25% of beliefs in 8 minutes
Why AI is effective: cool-headed, personalized, and fact-based
The double-edged sword: AI can also persuade people to believe lies
Transition to prevention: Sander van der Linden's approach
Introducing Prebunking: inoculating against misinformation
How prebunking works: exposing manipulation techniques
The vaccine analogy and using historical propaganda to teach
Instagram experiment: prebunking video boosts detection to 60%
The truth bias and the challenge of living in the gap
Conclusion: Learning to leave room for doubt
Credits and sponsor information
Listen ad-free on Castria