Intellectual humility is recognising that the confidence you feel and the probability you are right are separate quantities, and that the first is a poor estimate of the second. In practice it means holding a belief firmly enough to act on while keeping it revisable, which is harder than either certainty or blanket doubt. It has nothing to do with thinking less of your own intelligence.
This page also answers how you change your mind, because the two questions have one answer. Changing your mind is what intellectual humility looks like from the outside.
What the field agrees on, and what it does not
Researchers disagree about the boundaries of the construct, so it is worth separating the settled part. Porter and colleagues, reviewing the field across personality, judgement, education and organisational research, identify a metacognitive core carrying scholarly consensus: recognising the limits of one's knowledge, and being aware of one's fallibility. Graded entry.
Around that core sit social and behavioural features on which agreement is weaker: recognising that other people may hold legitimate beliefs different from your own, and being willing to reveal ignorance in order to learn.
The same review names the measurement problem, and it is a sharp one. Where intellectual humility is seen as desirable, in a job interview for instance, self-report questionnaires make a false impression easy to create. Which is why the rest of this page leans on the one setting that scores the thing rather than asking about it.
Where the trait has actually been scored
Most writing about open-mindedness is exhortation. Forecasting is the exception, because the answers resolve and the people can be ranked.
Tetlock and Gardner's account of the Good Judgment Project found accuracy over geopolitical questions to be measurable and learnable, and produced by a set of habits rather than by credentials or subject expertise. Superforecasting (2015). Four of those habits are the operational definition of the trait:
Confidence expressed in probabilities. Sixty per cent can be scored against what happened. Likely cannot, which is what makes it comfortable.
Updating in small increments, frequently. The accurate forecasters moved constantly by a few points. The inaccurate held a position and then reversed it wholesale, which looks like decisiveness and performs worse.
Actively seeking disconfirming evidence, rather than evaluating it fairly when it turns up.
Working in teams where challenge was expected, which converted disagreement from a threat into an input.
Notably, subject expertise without feedback performed poorly. That is the same boundary Kahneman and Klein drew for intuition: where the environment returns no clear result, confidence detaches from accuracy while continuing to grow. Graded entry.
Why people do not update
The obstacle is rarely ignorance of the counter-evidence. It is cost.
Julia Galef's contribution is the mechanism: people revise when revising costs less than defending, and the price of defending is set by how much of their identity the belief is carrying. The Scout Mindset (2021). Which means the work happens before the evidence arrives, not after. Once a position is publicly yours, the cost of abandoning it is fixed and no amount of good faith at the moment of challenge will lower it.
Adam Grant makes the same separation the centre of his account of rethinking, and adds the organisational version: teams that treat being wrong as a status loss will produce people who are never wrong out loud. Think Again (2021). Both are argued rather than measured, and both are consistent with what the scored forecasting data shows.
Changing your mind is not weak judgement
The common objection treats revision as inconsistency. It has the logic backwards. If a person's beliefs never move, either they were right about everything at the outset or their beliefs are not responding to evidence, and the second is far more likely.
What separates good updating from mere drift is whether the revision tracks new information or tracks social pressure. That distinction cannot be made from memory, because memory reconstructs past beliefs to match present ones and the feeling of having been broadly right survives almost any actual record. A dated written position is the only instrument that works. This estate keeps one: predictions including the misses, and corrections.
What AI changes
Nothing above involves a machine. One thing follows from the tools rather than from new evidence about people: a system that will produce a fluent, well-argued case for any position lowers the cost of defending whatever you already think.
Constructing a persuasive defence used to take effort, and effort is a tax on motivated reasoning. Galef's mechanism says people update when revising is cheaper than defending. Subsidising the defence side of that comparison moves the balance without anyone deciding to be less open-minded.
That is a claim about the price of arguments, not a measured effect on human belief revision, and it should be read that way. The use question follows directly: whether you ask the model for the strongest case against your position or the strongest case for it. Handled at how do I get AI to challenge me. And the prose arrives confident either way, which is a property of the writing rather than of the case behind it: why does AI sound so confident.
What this page does not establish
The forecasting evidence comes from questions with resolution dates. Most consequential judgements at work never resolve cleanly. That is the environment Kahneman and Klein identified as the one where confidence deserves least trust, and the same absence of scoring makes the habits above hardest to practise there.
Galef and Grant are argument rather than measurement. Their mechanisms are plausible, widely reported and not established by controlled evidence, and this page uses them for the human mechanism rather than as proof of an effect.
The construct itself is only partly settled. Porter's review finds consensus on the metacognitive core and continuing disagreement about what else belongs, and a person's score on a questionnaire is not the same as the trait, particularly where showing it is rewarded.
One half of this usually gets left out. Perpetual openness is its own failure, since a person who reopens every settled question never acts on anything. Calibration is the target, and it can be missed from either side.
Key sources
- Porter, T., Elnakouri, A., Meyers, E. A., Shibayama, T., Jayawickreme, E. and Grossmann, I. (2022). Predictors and consequences of intellectual humility. Nature Reviews Psychology, 1, 524-536. Graded entry.
- Tetlock, P. E. and Gardner, D. (2015). Superforecasting: The Art and Science of Prediction. Crown. In the essential works.
- Galef, J. (2021). The Scout Mindset: Why Some People See Things Clearly and Others Don't. Portfolio. In the essential works.
- Grant, A. (2021). Think Again: The Power of Knowing What You Don't Know. Viking. In the essential works.
- Kahneman, D. and Klein, G. (2009). Conditions for intuitive expertise: a failure to disagree. American Psychologist, 64(6), 515-526. Graded entry.
Related SuperSkills research
On the adjacent faculties, critical thinking and judgement. On knowing what you know, metacognition and the illusion of competence. On asking better, what makes a good question. On the record this estate keeps of its own calls, predictions and corrections.
About this research
Rahim Hirji is the author of SuperSkills (Kogan Page, 2026), keynote speaker on AI and human capability, and founder of The SuperSkills Intelligence Company. The scout framing is Galef's, the forecasting findings are Tetlock's, and neither is a SuperSkills coinage.
Cite this
Hirji, R. (2026). What is intellectual humility? The SuperSkills Intelligence Company. Last reviewed 30 August 2026. thesuperskills.com/research/what-is-intellectual-humility