Sean Carroll's Mindscape: Science, Society, Philosophy, Culture, Arts, and Ideas

333 | Gordon Pennycook on Unthinkingness, Conspiracies, and What to Do About Them

October 27, 2025

Key Takeaways Copied to clipboard!

  • Susceptibility to misinformation and pseudo-profound bullshit stems less from motivated reasoning and more from "unthinkingness," or relying too heavily on intuitive, effort-free cognitive processing. 
  • People who believe in conspiracy theories often exhibit general overconfidence, demonstrated by a tendency to overestimate how much others agree with their fringe beliefs (a large false consensus effect). 
  • Conversations with patient, fact-providing AI chatbots can significantly decrease a person's confidence in their conspiracy beliefs, suggesting that evidence and engagement can be effective tools against misinformation. 
  • People susceptible to conspiracy theories are often those who desire dialogue and are interested in exploring topics, suggesting their susceptibility is often a failure of science communication rather than a desire to believe falsehoods. 
  • Conversations with AI chatbots designed to counter specific conspiracy theories can lead to sustained belief changes, with effects lasting for at least two months without decay. 
  • Combating misinformation effectively requires getting high-quality information into the market and ensuring people engage with it, as the problem lies more with the information environment than with inherent human motivations to be irrational. 

Segments

Introduction to Unthinkingness
Copied to clipboard!
(00:00:00)
  • Key Takeaway: False beliefs are often nurtured by social groups and are explained by “unthinkingness” rather than solely by cognitive biases or motivated reasoning.
  • Summary: The episode introduces the concept that people often hold false beliefs, sometimes reinforced by their social circles. Guest Gordon Pennycook argues that susceptibility to misinformation and pseudo-profound bullshit is primarily due to a lack of careful, reflective evaluation of claims. This lack of critical engagement is termed “unthinkingness.”
Pseudo-Profound Bullshit Research
Copied to clipboard!
(00:05:11)
  • Key Takeaway: The Ig Nobel Prize-winning research on pseudo-profound bullshit measured receptivity to statements that sound profound but lack meaning, often correlating with reliance on intuition.
  • Summary: Gordon Pennycook discusses his Ig Nobel Prize-winning work on pseudo-profound bullshit, exemplified by sentences generated from Deepak Chopra’s buzzwords. People who rely more on intuition and gut feelings rate these meaningless statements as more profound. Bullshit, in this context, is defined by a disregard for truth or evidence, contrasting with lying where the truth is actively subverted.
Intuition Versus Deliberation
Copied to clipboard!
(00:11:53)
  • Key Takeaway: Cognitive processing involves both fast, intuitive responses and slower, effortful deliberation, and falling prey to bullshit occurs when relying too much on the former.
  • Summary: The mind processes information through automatic, intuitive responses (System 1) and effortful deliberation (System 2), though these are interconnected processes, not separate brain parts. Over-reliance on intuition leads to accepting bullshit, while analytical thinking helps distinguish profundity from nonsense. People vary in their disposition to engage in effortful deliberation.
Overconfidence and Conspiracy Beliefs
Copied to clipboard!
(00:41:23)
  • Key Takeaway: Conspiracy believers are more likely to exhibit general overconfidence, believing they are correct even in novel, unrelated tasks where performance is random.
  • Summary: General overconfidence, distinct from the Dunning-Kruger effect, correlates with susceptibility to conspiracy theories. Overconfident individuals overestimate their performance on novel tasks where success is pure chance. This overconfidence is linked to being more intuitive and less likely to question initial impressions, such as giving an immediate, incorrect answer to a simple logic puzzle.
False Consensus and AI Debunking
Copied to clipboard!
(00:48:48)
  • Key Takeaway: Conspiracy believers strongly overestimate public agreement with their views (a massive false consensus effect), but they are receptive to evidence when presented by patient AI chatbots.
  • Summary: Conspiracy believers often believe they are in the majority, drastically overestimating public agreement with their views due to echo chambers and social dynamics. Research shows that conversations with AI chatbots, which provide detailed counter-evidence patiently, can significantly decrease believers’ confidence in their conspiracy theories. The effectiveness of this intervention relies on the facts and evidence provided, not merely the AI’s politeness.
Conspiracy Theorists’ Desire to Talk
Copied to clipboard!
(00:57:25)
  • Key Takeaway: Conspiracy theorists are surprisingly interested in dialogue rather than just confrontation.
  • Summary: People susceptible to conspiracies want to discuss their beliefs, suggesting they are not inherently opposed to dialogue. This interest indicates a potential failure in science communication, where they pursued conspiracy rabbit holes instead of scientific ones. Reality is harder to make interesting than constructed narratives, posing a challenge for factual communication.
Motivation Theory Critique
Copied to clipboard!
(00:58:05)
  • Key Takeaway: The theory that conspiracy believers are fine with falsehoods because they ‘want to’ is flawed because it only describes ’the other,’ ignoring that everyone has biases.
  • Summary: The general viewpoint that conspiracy theorists enjoy believing falsehoods is problematic because no one believes they want to be wrong; everyone is susceptible to their own biased rabbit holes. People generally want to reason, talk, and are susceptible to evidence, even if the current information environment obscures this fact. Losing the misinformation war is attributed more to the information environment than to the people themselves.
AI Efficacy and Overconfidence
Copied to clipboard!
(00:59:16)
  • Key Takeaway: Overconfident individuals are less willing to invest effort in seeking counter-evidence from chatbots, though reflectivity is a stronger predictor of change.
  • Summary: The strength of counter-evidence from chatbots correlates more strongly with how much a person values evidence and their reflectiveness than with overconfidence alone. Overconfident people may be less willing to put in the effort required to receive strong counter-evidence from the AI interaction. Intuitive thinkers can experience epiphanies when confronted with simple counter-evidence, such as realizing steel beams lose carrying capacity when heated.
Long-Term AI Impact and Carryover
Copied to clipboard!
(01:00:26)
  • Key Takeaway: Belief changes induced by AI conversations persist long-term and can create small, generalized skepticism toward other unrelated conspiracy theories.
  • Summary: Follow-up contact one and two months later showed that the effect of changing minds via AI conversation did not decay. This sustained change is unusual in psychological experiments, indicating a genuine shift in belief rather than temporary compliance. Furthermore, successfully debunking one conspiracy led to a small but measurable reduction in belief in other, unconnected conspiracy theories, suggesting a generalized increase in skepticism.
AI Accuracy and Training Constraints
Copied to clipboard!
(01:02:45)
  • Key Takeaway: The AI’s accuracy in debunking conspiracies stems from its training on the vast, fact-checked data of the internet, constrained by hard-coded source quality parameters.
  • Summary: The specific AI used in the study was highly accurate, with external fact-checkers confirming nearly all claims made against the conspiracy theories. This accuracy is tied to the task being perfectly suited to the AI’s training data, which reflects the majority view of the internet. Attempts to force AI like Grok to espouse fringe views often fail because reality and hard-coded source quality checks act as constraints.
Lessons for Combating Misinformation
Copied to clipboard!
(01:05:32)
  • Key Takeaway: The primary solution to misinformation is overwhelming the environment with good information, countering the idea that truth is irrelevant due to identity-driven motivations.
  • Summary: The research suggests that the truth does matter, contradicting theories that identity and bias completely override factual engagement. The battle against misinformation is fundamentally about ensuring good information is widely available to overwhelm bad information in the current diffuse media landscape. Educators and science communicators are performing the necessary work by getting information out, even though this task is inherently difficult.