Thinking Under Uncertainty

In the first two posts of this series, I focused on judgment and sensemaking. Judgment helps students choose among possibilities. Sensemaking helps them organize information into understanding.

But both of those forms of thinking become more difficult when the situation is uncertain. And uncertainty is where much of real thinking happens.

In education, we often try to make learning clean and structured. We give students clear prompts, defined instructions, grading rubrics, examples, and expected outcomes. Those supports matter. Students need structure. But the world students are preparing to enter is rarely that neat.

In professional life, civic life, and personal decision-making, people often have to act without complete information. They make decisions when the evidence is incomplete. They weigh options when every choice involves trade-offs. They interpret situations where the answer depends on context. They revise their thinking when new information appears.

That is thinking under uncertainty.

And it may become one of the most important forms of human thinking in an AI-rich world.

AI can create the impression that uncertainty has disappeared. Ask a question, and an answer appears. Ask for a recommendation, and the tool offers one. Ask for an explanation, and the response often arrives with clarity, structure, and confidence.

That confidence can be useful. But it can also be misleading.

Because many of the most important questions students will face do not have simple, final, or universally correct answers. What is the best solution for this situation? Which source should I trust? What matters most in this decision? What risks should I consider? What information is missing? What would change my conclusion? How confident should I be?

Those questions do not ask students simply to retrieve information. They ask students to reason through uncertainty.

That distinction matters.

When students are only asked questions with clear answers, they may come to see learning as the process of finding the correct response. That kind of learning has value, especially when students are building foundational knowledge. There are facts, concepts, procedures, and principles students need to know.

But if learning stops there, students may not develop the capacity to think when the answer is unclear. And increasingly, that is where human judgment matters most.

AI can generate a confident response to an uncertain question. But confidence is not the same thing as certainty. A response can sound complete while leaving out context. It can recommend a course of action without understanding local constraints. It can summarize evidence without recognizing what evidence is missing. It can present one interpretation as if it is the only reasonable one.

Students need to be able to notice that.

They need to recognize when a situation requires caution. They need to know when to ask for more information. They need to identify assumptions, weigh probabilities, consider alternatives, and decide how much confidence a conclusion deserves.

Those are not just academic skills. They are life skills.

In a healthcare setting, uncertainty might involve interpreting symptoms that could point in multiple directions.

In business, uncertainty might involve making a decision with incomplete market data.

In education, uncertainty might involve deciding how to support a student when the cause of the struggle is not immediately clear.

In technology, uncertainty might involve troubleshooting a system when several possible causes seem plausible.

In civic life, uncertainty might involve evaluating competing claims when no single source provides the full picture.

Across contexts, uncertainty does not mean thinking has failed. It means thinking is required.

That is an important message for students. Many students are uncomfortable with uncertainty because they have been trained to look for the answer. When the answer is not obvious, they may assume they are doing something wrong. They may feel frustrated, anxious, or unsure how to proceed.

AI can intensify that discomfort because it often provides immediate closure. It offers an answer before students have had time to wrestle with the problem. That can be helpful when students are truly stuck. But it can also short-circuit the learning process if students accept the first response as the final answer.

Sometimes learning requires staying with uncertainty a little longer.

·      It requires asking, “What do I know?” and “What do I not know yet?”

·      It requires separating evidence from assumption.

·      It requires considering more than one possible explanation.

·      It requires being willing to revise a conclusion.

Those are habits students need to practice.

One way educators can support this is by designing assignments that include ambiguity on purpose. Not confusion. Not unclear directions. But meaningful ambiguity that asks students to reason, decide, and justify.

For example, instead of asking students to identify the single correct solution, we might give them a scenario with incomplete information and ask them to propose a reasonable next step. What would they do first? What evidence supports that choice? What additional information would they want? What risks are involved?

That kind of prompt helps students practice thinking under uncertainty.

In a writing course, students might compare two sources that offer different interpretations of the same issue. Rather than asking which one is simply “right,” we might ask students to evaluate the strength of each argument, identify what evidence is missing, and explain where their confidence is strongest or weakest.

In a technical course, students might troubleshoot a problem where multiple causes are possible. Instead of grading only the final fix, we can ask students to explain their diagnostic process. What did they test first? Why? What did they rule out? What would they try next?

In a discussion, we might ask students to respond to a question where reasonable people could disagree. The goal would not be to force consensus. The goal would be to help students explain their reasoning, listen to alternative perspectives, and revise their thinking when appropriate.

These kinds of activities help students understand that uncertainty is not something to avoid. It is something to work through.

AI can be part of that process, but only if we use it carefully.

Students might ask AI to generate multiple possible explanations for a problem. Then their task could be to evaluate which explanation is most plausible and what evidence would be needed to support it.

They might ask AI for a recommendation and then identify what assumptions the recommendation appears to make.

They might ask AI to argue the opposite side of a conclusion and then reflect on whether their own thinking changed.

They might ask AI what information would be needed before making a decision and then compare that list to their own.

In each case, AI is not replacing the student’s thinking. It is creating material for the student to evaluate.

That distinction is critical.

Because the goal is not to teach students that AI has the answer to uncertainty. The goal is to help students use AI without surrendering their responsibility to think.

Students need to learn that a generated answer is often the beginning of thinking, not the end of it.

This is especially important because uncertainty can make people vulnerable to overconfidence. When a tool provides a clear answer in an unclear situation, the clarity can feel comforting. It reduces the discomfort of not knowing. But sometimes that discomfort is useful. It reminds us to be careful. It reminds us to ask more questions. It reminds us that our first conclusion may not be the best one.

Educators can help students develop a healthier relationship with uncertainty by making it visible in our own thinking.

We can say things like:

·      Here is what we know.

·      Here is what we do not know yet.

·      Here are the assumptions we are making.

·      Here are two possible interpretations.

·      Here is why I would choose this option for now.

·      Here is what might change my mind.

When students hear that kind of reasoning, they begin to see that uncertainty is not a weakness. It is part of serious thinking.

And when students practice that reasoning themselves, they become better prepared for the kinds of problems AI cannot simply solve for them.

Because AI can generate possible answers. But students still need to decide how much confidence those answers deserve.

They need to know when to trust, when to question, when to seek more evidence, and when to revise their thinking.

That is thinking under uncertainty.

And in a world where answers are increasingly easy to generate, the ability to reason carefully without complete certainty may become one of the most important human capacities we can help students develop.

Continuing the Conversation

Series 1: AI Is Exposing Existing Problems ✓ Completed
Series 2: What We Do About It ✓ Completed
Series 3: Cultivating Human Thinking
Current Post (3 of 8): Thinking Under Uncertainty
Next Up: Curiosity: Asking Better Questions