We often assume that confidence comes with competence. But sometimes we are overconfident in what we know and, at other times, we underestimate what we can actually do. In a world of rapid change and increasingly sycophantic AI, students will need to recognize the limits of their own knowledge while remaining confident in their ability to learn. This is the power of confident humility, and it might be one of the most important mindsets for navigating an uncertain future.

Humility without Confidence

I’m leading a workshop on student agency with a group of secondary math teachers. I’m blown away at the depth of their content knowledge and their willingness to wade into the nuance of pedagogy. The day is going so well. But then, I share the short assignment immediately room grows tense. Two participants get up to use the restroom. I can sense from the collective body language that this is the last thing anyone wants to do right now. 

“I am going to ask you to share with a partner but I’m not grading it, I promise. This is a workshop.” 

A few teachers nod. One man says, “You know we’re math teachers, right?” 

“I do.” 

“It doesn’t need to be artistic? You sure. Because I’m not an artist,” a woman says. 

I nod. “Just get your ideas out visually without using any words. The goal is the process. We want to make it visual so that we can remember it.” 

“I’m not a visual learner,” someone says. 

“There’s no such thing as learning styles,” another teacher corrects. “We all learn visually. It’s just that some of us don’t like to draw.” 

“Remember. It doesn’t have to look perfect. It doesn’t even have to look good. It’s just a short exercise,” I say. “Go ahead and begin.” 

I wander the room as they begin their visual reflections. Many of them scroll their phones for ideas. A few use AI to get a sense of how to draw what they want to draw. Several teachers are slow to get started. A few of them look at other members of their table to get a sense of what to do (despite the fact that I provided some exemplars). I watch one teacher sketch things out quickly and turn the paper over so no one sees it. 

When we are finished, as they explain their visual reflections to a partner, I am struck by the mismatch between most people’s skill levels and their perception of skill levels. Every person’s doodles are unique and interesting and, honestly, pretty good. They aren’t professional illustrators but they are better at communicating visually than they realize. 

Part of this is that we sometimes mistake experience for expertise. Everyone (myself included) overestimates how well they drive and underestimates how well they draw. In this moment, they are experiencing intellectual humility without a sense of confidence. The result is a certain fear and anxiety that keeps them from getting started and leads to many of them failing to get started or struggling to try hard. 

 

Confidence Without Humility

I contrast this to a meeting I had with a group of professors, literacy experts, and parent volunteers. Our focus was on the science of reading and early literacy. I am not not an early literacy person by any means. However, I taught reading intervention to middle schoolers and I have taken a deep dive into the journal articles to try to understand the complexity of literacy.

“I think it’s clear that schools aren’t teaching phonics and that’s at the heart of our literacy crisis,” a parent advocate begins. She then tells the story of her own child who is dyslexic and her experience in advocacy. Members at the table shake their head. 

One by one, people confirm her stories with stories of their own. A few reference their awakening to advocacy when they heard Sold a Story. A few researchers offer a critique of whole language approaches that seems, on the surface, to be a straw man version of whole language. 

The message is clear. We know the problem. Kids aren’t learning phonics. We have the solutions. More phonics, more standards, and even a required set of courses at the secondary level for early literacy. That’s right. Let’s get high school PE and Calculus teachers learning early literacy phonics instruction. 

Eventually, I offer some questions. “As someone who taught English Learners, I often see interference between their first language and English. Spanish, for example, is truly phonetic in its spelling. English isn’t. We have multiple . . .” 

“How can you say that?” Someone interrupts. “We just have a more complicated set of phonics. That’s why it’s so important.” 

I try later to bring up other factors that are critical, such as reading endurance, but I’m told the reason kids don’t have reading endurance isn’t the distraction epidemic or short attention spans so much as the cognitive load of memorizing sight words and reading slow. I mention the role of complex verb tenses, prior knowledge, and other factors that have led to lower reading levels at the middle school level. But again, I’m told that I simply don’t know enough about phonics.

To be clear, I am a proponent of the science of reading. I see the validity in the critique about how we teach phonics at the early elementary level. I have never been a whole language proponent. But as someone who delves into the science of learning, I want to approach a challenge with big questions that lead to complexity and nuance.

I leave this meeting realizing that many people in this room have confidence without humility. Due to their powerful personal experiences, it’s hard to see that something as complex as literacy might have multiple challenges that need to be addressed. And when everyone in the room is on the same page, it’s hard to admit that there might be something valid on another page. 

To be honest, I’ve been there before with reading. When I first taught reading intervention, I assumed that the issue was motivation. If I could just get kids to fall in love with reading, all would be great. True, I’d need to do some phonics clean up work and maybe practice some reading strategies. But then I learned that there were students who desperately wanted to read but couldn’t. I then discovered that the challenges were often multifaceted and the targeted interventions had to be structured and personalized. Did motivation play a role? Absolutely. But I had oversimplified the solution. 

As I think about these two different interactions I had, I am reminded of the relationship between confidence and humility. We see this with knowledge and expertise in understanding concepts (the science of reading experts in the meeting) and in mastering skills (the math teachers drawing). 

 

The Dunning Kruger Effect

In 1999, David Dunning and Justin Kruger published their groundbreaking study Unskilled and Unaware of It. They chose multiple domains (ranging from humor to logical reasoning to grammar) and discovered an alarming trend. Often the least competent performers were actually the most confident. So, you end up with a mismatch on your skill level and your perception of skill level. 

In this stage, we can be wrong without having the prerequisite knowledge to realize that we are wrong. We can be unskilled but blissfully unaware of our lack of skill.

In another study, they demonstrated that there’s often a lag that occurs toward the beginning of skill development where one’s skills are actually higher than their perception. This leads to a sense of imposter syndrome, where you feel that you aren’t good at something but you are actually far more competent than you thought. 

When I think about the math teachers doing a visual reflection, they were in the imposter syndrome space. Their skill level was far higher than their perceived skill level. In this moment, they lacked confidence but had humility. 

However, when I think about the second group, I feel like there was an overconfidence in the conceptual understanding of reading and the systems at work. Personal experiences, as powerful as they may be, are not always normative. And policies need to address complex contextual factors. This group had confidence. I loved their confidence in their convictions. I just wish they had tempered this confidence with humility.  

Some nuance here . . . 

Over the last quarter century, this research has grown oversimplified. For example, I have seen countless “Mount Stupid” diagrams that oversimplify the research in a way that misses the nuances of the research. The original studies brought up important ideas about context (where you apply skills matter) and the challenges in multivariate skills (often a “skill” is actually a combination of multiple skills applied in tandem). 

Some critics have actually doubted the validity of the original research, instead claiming that they discovered a natural regression toward the mean. Others have called for a deeper analysis of demographic variables (are certain groups more likely to be overconfident?) I would also point out that overconfidence in ideas or conceptual understanding might be different from overconfidence in skills. 

Still others have added another layer of nuance where skill development can actually increase self-efficacy to the point of overconfidence. We see this when we drive. Most of us grow overconfident to the point of carelessness when we move into autopilot in driving near our homes and are thus in the danger zone for accidents. 

This nuance actually reinforces a key idea here. Even when thinking about the research on overconfidence, we should approach things with a hefty dose of intellectual humility. 

 

What is Confident Humility?

I first heard the term “confident humility” in Adam Grant’s book Think Again. Grant suggests that we should approach our beliefs like a scientist who forms hypotheses, tests assumption, examines evidence, and remains willing to revise conclusions. A critical element here is curiosity. Instead of looking for evidence to back up presuppositions (which can lead to overconfidence and confirmation bias) you openly ask for ideas that might contradict your thinking. 

With confident humility, you remain confident in your ability to learn while remaining aware that your current knowledge may be incomplete. Humility without confidence often leads to anxiety and risk-aversion. We should absolutely feel confident about concepts we’ve learned, knowledge we’ve attained, and skills that we’ve mastered. But that sense of mastery should be tempered with a humble recognition that there might be new ideas or better approaches that we might be missing. 

The intellectually humble person can make a strong argument while simultaneously asking, “What would cause me to change my mind?” They can perform at a high level and still ask, “What can I do better next time?” 

I feel like confident humility is a critical mindset that students will need as they navigate the maze of an uncertain future. 

 

Why This is Critical in an Uncertain Age

For years, we were taught a consistent formula. Work hard. Graduate from college. Climb the corporate ladder. We still see this formula embedded in our school system. Take the AP courses. Go to the right university. Pick the perfect degree. Then go find the career that provides what you need in life. 

Note that this formula was never equally accessible to all people. Women and gender minorities faced big hurdles and outright roadblocks along the way. Racism and systemic injustice meant the door was shut to certain segments of our population as well. Still, this was the formula that we grew up following. 

However, the ladder is currently gone and in its place is a maze of uncertainty. With generative AI and automation, we don’t know what our jobs will look like in the future. With our current changes in supply chains and the disruptions in social and economic systems, it’s hard to predict the future of nearly any industry. 

If the last decade has taught us anything it’s that the only thing we can predict is unpredictability. 

Sometimes I wish we had a predictable ladder. I wish there was a blueprint or a roadmap or a set of directions. But that’s not our current (or future) reality. Our students are going to change roles, industries, and professions in a way that we’ve never seen before. They will repeatedly find themselves moving from expert to novice to somewhere in between. 

But that’s why confident humility is so vital. With a ladder, it’s important to know the answers. With a maze, it’s important to know the questions. You need to know what you don’t know and learn how to pivot and iterate as you go. 

 

Confident Humility is Vital in a Generative Age

For the last 50-60 years, we have seen countless pop culture stories of robots who go rogue. SkyNet comes to mind as the most nefarious example of this combination of surveillance with dark AI tendencies. It’s no surprise, then, that our computer scientists and software engineers have worked toward designing inherently pro-social AI machines. We want our generative AI to lean toward being positive and pro-human. That’s a good thing. 

And yet, in the process of designing positive generative AI, we now face a new challenge: the AI Sycophant who is all too eager at finishing our half-baked ideas and telling us how awesome we are. On some level, working with a chatbot can feel like being stuck in the Lego movie. Everything is awesome. Always. Every. Single. Time. 

In a recent Science article, researchers examined eleven generative AI models and found that AI affirmed the users’ actions significantly more than humans ever would. But here’s the scary part. This consistent sycophantic response process changed our own behavior as well. People became convinced that their convictions were right. They were less likely to change their minds. They were less willing to take responsibility for personal conflicts. 

So, if you think about a child using AI, they are more likely to get a chatbot who agrees with their thinking process without offering any pushback or critical feedback. In many cases, the AI is also highly capable, which means the chatbot can take a half-baked idea and make it sound coherent and authoritative. 

The net result is an overconfidence loop. I have an idea. The AI validates my idea. My confidence grows. The AI strengthens my argument and even fills in the gap. My argument now feels not only correct but more sophisticated. I then grow more confident in the original idea. In the meantime, there is no friction. Nobody challenges my idea. 

Note that the AI doesn’t create overconfidence. We are plenty good at that as humans. But it does amplify the overconfidence without offering any friction. So, instead of functioning as a thought partner, it becomes a veritable yes-man offering sycophantic responses to our worst ideas. 

As I think about this next generation, I worry about the role of AI overconfidence. It’s easy to say, “Just teach students to think critically about AI. Ask them to question it. Get them in the habit of asking, ‘Could I be wrong about this?’” But here’s the thing. That requires intellectual humility. And if students are interacting frequently with sycophantic AI chabots, there’s a very real danger in children not even seeing this bias toward the positive. 

So, at a time when intellectual humility is more vital than ever, our students will be interacting on a regular basis with a technology platform that facilitates overconfidence. 

 

What does this look like in the classroom?

For the last two decades, we have shift in education toward standardized testing. We had No Child Left Behind and Race to the Top which both emphasized growth and achievement on multiple choice tests. But this approach also pushed a core idea into the hidden curriculum. It was an underlying message of certainty. 

Don’t question answers. Answer questions. Don’t take take too long to figure it out. We’re on a race to the tippy top. You don’t want to be left behind. 

In fact, as a teacher I was told students should not go back to previous questions on their test and change their answers. Research actually suggests the opposite. Second-guessing your answers actually improves test scores. But I didn’t know it at the time. 

So, students end up valuing certainty over uncertainty. They raise their hands when they know the answer. They take quizzes that distinguish correct answers from incorrect ones. They write essays focused on defending a thesis with facts and argumentation. We have debates that reward the person who makes the strongest case. 

There is value in all of these things. But students also need opportunities to practice changing their minds. As I think about the maze of an uncertain future, our students will need to approach things with confident humility. We don’t simply need lifelong learners. We need lifelong unlearners and relearners.  

 

1. Teach Students to Calibrate Their Confidence

With this approach, we can have students compare their confidence level with their actual understanding. Before answering a question, students might indicate:

• I know this.
• I think I know this.
• I’m unsure about this.
• I don’t know this yet.

Then compare confidence with actual understanding. So, they might do the confidence level survey and then complete a concept map. Afterward, they compare their conceptual understanding to other concept maps. They might use this before an ungraded test and then compare their performance to their perceived skill level. 

We can also help students understand the emotions they might feel around this recalibration process. In my experience, students tend to do pretty well in addressing conceptual misunderstandings because these are often impersonal. So, if you are failing to understand the difference between macro and micro economics or if you are still developing an understanding of Newton’s Laws, fixing misconceptions can feel frustrating at times but also a bit of a relief. 

However, when you are calibrating your confidence in skill development, it feels more personal. You’re essentially answering the question, “Am I good at this?” And that can feel scary for students (and, if I’m being honest, for adults as well). We want to know if we are okay. 

This is where it helps to frame skill development with a growth mindset, where you focus on rewarding effort but then reflecting on performance. 

The conversation becomes, “Is there a different way to do ______?” Or even “How do I take _______ to the next level?” 

I used to work with a teacher who did “epic fails” at the end of each day. Students would share, “My epic fail was ______. What I learned was _______. Next time I will ______.” They would share with a partner and then he’d call on a student randomly to share with the whole class. Was this a bit cheesy? Perhaps. But did it promote positive risk-taking and confident humility? Absolutely. 

Confident humility can feel even more personal when we get into beliefs. If I think about the science of reading advocates, they have profound personal experiences that often connect to traumatic events. It’s no wonder they might skew toward overconfidence in their beliefs and potential solutions. This is where it helps to ask people to explain their beliefs (I call it the “Tell Me More” approach based on the musical Grease). 

 

2. Incorporate Revision into the Process

Here, we can ask students to document how their thinking evolved rather than hiding their initial misconceptions. For example, you might do a Wonder Sparks activity where students have an overarching interesting question like “Why don’t jellyfish sting themselves?” From there, they make a hypothesis and read an informational text. Afterward, they share why they were right or wrong by using the text to defend their claims. Note here that instead of making a claim and defending it (an argumentative essay) they are essentially saying, “Here is what I thought and why I was wrong.” They then end with their lingering questions. 

We can prompts such as:

• I used to think _____, but now I think _____.
• One assumption I had was _____.
• The evidence that challenged my thinking was _____.
• I’m still uncertain about _____.

If the focus is on skill development, we can have students engage in peer assessment where they actively seek out feedback. One of my favorite approaches is the twenty-minute peer feedback process. 

We can also help students learn to ask the right kinds of questions that lead to better revision. Adam Grant points out that when you want feedback on a performance asking, “What can I do differently next time?” (An advice based question) works better than “How did I do?” When it’s a conceptual understanding you might say, “What are some areas that I might be misunderstanding?” When getting feedback on a product, you might say, “What can I do to improve the _____ of _______.” 

Admittedly, this can be a challenge when we have limited time and packed curriculum maps. But when we can build revision into the process, we are also facilitating confident humility.

 

3. Teach Students to Seek Disconfirmation

Ultimately, we want students to seek out disconfirmation as a habit. We can do this with subtle tweaks. For example, we might shift from constantly using, “What evidence supports your claim?” To also asking”What is the strongest evidence against your claim?”

I love doing an activity where you have students create a utopia and then explain how this utopia becomes a dystopian reality. 

But you can do the same thing with nearly every idea by saying, “Imagine you were totally wrong about _______. Explain where you went wrong in your thinking.” For example, I love project-based learning but when I asked myself the question, “What if PBL is an awful strategy? What if it doesn’t work? Make the case against it.” This then led me to add more structure to the process and to design PBL in a way that reduced cognitive load. It offered a critical nuance I was missing. 

When students create projects, you might have them do a pre-mortem where they pretend that their prototype completely failed and then address the reasons for failure before revising their initial plan. 

If we think about the role of AI, this means reframing our prompting to focus on disconfirmation. 

Instead of :Here’s my argument. Help me make it stronger” we can say “Here’s my argument. What might I be missing?” Or “What is the strongest argument against my position?” Instead of saying “What do you think of ______?” (Which often leads to sycophantic responses) you can say “Which assumptions and I making that might not be true?” 

AI can reinforce overconfidence, but students can also learn to use it deliberately as a tool for intellectual friction by focusing on disconfirmation. 

 

4. Incorporate curiosity

There’s a great scene from Ted Lasso where he is playing darts with an arch-nemesis and he ends up telling the story of seeing a quote from Helen Keller stating, “Be curious, not judgmental.” 

I was curious about that quote. It just didn’t seem to be the kind of thing Helen Keller would have actually said. It turns out, it’s harder to pin down exactly who said this first. 

However . . . .

I think there’s some real validity in being curious and not judgmental. Curiosity is a critical component of confident humility. When you’re curious, you’re willing to re-examine ideas and approaches. You look at information and instead of jumping straight to analysis (where you are likely to fall victim to bias) you ask questions. I’ve already alluded to this with the example of a Wonder Spark activity. But you can also include moments of curiosity throughout your lessons. Students can ask questions after reading rather than as well as answering questions. You can take a break during direct instruction and have them ask questions and answer those as well.