Key points
- Access to AI answers reduces willingness to admit “I don’t know,” even when AI is wrong.
- Intellectual humility shields against believing AI-generated misinformation.
- Epistemic resilience—tolerating uncertainty—protects judgment and fosters ethical reflection.
New research shows that people with access to AI-generated answers are less willing to admit “I don’t know,” even when those answers are wrong
In the study, researchers asked participants very specific, obscure questions about films. Participants could always decline to answer and admit they did not know, suspending judgment instead of guessing. The stakes were raised with a monetary penalty for mistakes. But across every version of the experiment, having access to AI advice nearly eliminated participants’ willingness to say they did not know, even though AI advice was usually wrong on these questions
In another study, researchers found that intellectual humility, the metacognitive awareness of one’s own limited knowledge, acted as a protective cognitive filter against AI-generated health misinformation. Participants with higher intellectual humility rated pseudoscientific content as less credible
The Known Problem of Automation Bias
Longstanding evidence on automation biasillustratesthe human tendency to over-rely on and defer to automated outputs. This has been a concern in medical decision-making and other fields like ae of their limitations, and may incorrectly perceive and treat automated systems as infallible “like a calculator” or like GPS
Large language models (LLMs) tend to answer in a fluent, authoritative tone independent of whether the underlying information is reliable. This is highly persuasive, and there is not yet a direct way to gauge the level of certainty behind each answer without independently cross-checking it
This is further complicated by the fact that LLMs frequently do provide accurate and helpful information and resources, even though the way they arrive at that information fundamentally differs from how humans vet information. This has been termed epistemia(episteme is Greek for knowledge),which Walter Quattrociocchi and colleagues define as “a structural situation in which linguistic plausibility substitutes for epistemic evaluation, producing the feeling of knowing without the labor of judgment.”
When Should We Apply Our Judgment?
While reliance on AI for film questions does not have high stakes, it points to larger questions about our deference toward AI-generated information. I am seeing more patients bringing in AI chatbot opinions of their medical and psychological issues. Some of the information is indeed accurate and useful. But some of it does not quite fit their situations, especially where interpersonal or social judgment is involved
Wharton researchers Shaw and Nave have named this risk of deferring to AI outputs as “cognitive surrender,” the concern that people will adopt AI outputs with minimal scrutiny. The trouble is that the devil is in the details
Applying our own scrutiny and judgment takes effort—and it is not always clear when and how often we should be expending this effort. This challenge is documented in self-driving cars. The levels of driving automation require different degrees of human oversight, and the riskiest zones are the middle levels, where a person is asked to supervise a system that is mostly right and the need to step in is rare. The constant vigilance required and sustained monitoring leads to a decline in attention, typically within 15 to 30 minutes, and cognitive fatigue (”vigilance decrement”) is linked to delayed reaction times.
Valuing Our Capacity for Uncertainty
There is a broader issue with the erosion of our ability to hold uncertainty, a capacity British psychoanalyst Wilfred Bion pointed out as essential, citing poet John Keats’s term negative capability, the ability to tolerate uncertainty, mystery, doubt, and frustration without rushing to a premature conclusion or “knowing” (Bion, 1970)
As a therapist, I have found that helping people through some of life’s most difficult situations often includes helping them expand their capacity to sit with uncertainty—whether with career, relationship, moral, or existential questions, the anxiety of facing illness, or the inevitable pain of loss and grief. I see this as a form of epistemic resilience—one that is deeply worth cultivating to help us move through difficulty
The ability to hold space and resist rushing to judgment is also integral for ethical and moral reflection. Ethicist Sylvie Delacroix argues that “productive uncertainty rather than efficient resolution” is essential to the infrastructure of ethical transformation. Prematurely foreclosing ethical questions, which can occur when we overtrust and defer to AI responses, prevents us from holding space for uncertainty. We need this space to make difficult discussions generative
Artificial Intelligence Essential Reads
Double Alignment for a Healthy Relationship With AI
When Should AI Enter the Classroom?
The Strengths of Epistemic Resilience
In my practice, I am seeing people express immediate relief, excitement, and even thrill at the expedience of LLM answers, which are quick, concrete, and well-defined. Sometimes it echoes the excitement of getting answers from a psychic or astrologer who offers a blueprint for the future. Both can reduce the short-term anxiety of uncertainty but can come at the cost of narrowing one’s own imagination and autonomy. Uncertainty is uncomfortable, but it also gives us the capacity for superposition, to hold multiple possibilities at the same time. Our capacity to hold uncertainty allows for an openness toward our unknown future, enhancing our agency and self-determination.
There can also be more insidious relational consequences in automatically deferring to AI. Imagine two partners in the middle of a disagreement over a nuanced quandary. One consults their AI chatbot, asking it to opine. This seems like a neutral third-party, but it is not. This can also result in feelings of betrayal, anger, loss of trust, and disrupt the mutual feeling of being on the same team, by implying there is more trust in AI than in each other. Repeatedly outsourcing to AI can short-circuit growth through communication and conflict resolution.
Nurturing wonder, discomfort, and reflection, without rushing to judgment or conclusions, takes real effort. All this does not mean that we should avoid integrating or relying on helpful AI systems. But in hybrid human-AI systems, it is important to consider not only short-term ease but also the long-term benefits of cultivating our capacity to handle uncertainty and epistemic humility
Copyright Marlynn Wei, MD, JD © Copyright 2026. All Rights Reserved
Delacroix, Sylvie. “Structural Uncertainty and the Conditions of Transformative Agency.” Topoi, ahead of print, March 12, 2026. https://doi.org/10.1007/s11245-026-10392-8
Marcoccia, Chiara, Walter Quattrociocchi, and Valerio Capraro. “AI Advice Suppresses People’s Willingness to Say “I Don’t Know”, Even When the Advice Is Wrong and Accuracy Is Incentivized”. PsyArXiv, July 15, 2026. osf.io/preprints/psyarxiv/5y6m4_v1
Quattrociocchi, Walter, Valerio Capraro, and Matjaž Perc. “Epistemological Fault Lines Between Human and Artificial Intelligence.” arXiv.Org, December 22, 2025. https://arxiv.org/abs/2512.19466v1
Rządeczka, Marcin, Maciej Wodziński, Kacper Zacharski, and Marcin Moskalewicz. “Intellectual Humility as a Cognitive Filter for AI-Generated Health Misinformation. An Evolutionary Perspective on Epistemic Vigilance.” arXiv:2606.03377. Preprint, arXiv, June 10, 2026. https://doi.org/10.48550/arXiv.2606.03377
Shaw, Steven D and Nave, Gideon, Thinking—Fast, Slow, and Artificial: How AI is Reshaping Human Reasoning and the Rise of Cognitive Surrender (January 11, 2026). https://doi.org/10.31234/osf.io/yk25n_v1, The Wharton School Research Paper , Available at SSRN: https://ssrn.com/abstract=6097646 or http://dx.doi.org/10.2139/ssrn.6097646
