By Danielle Wallace
If AI takes more of the routine work off our plates, what happens next? The promise is that people will spend more time solving harder problems, exercising critical thinking, and making the kinds of judgments that still require human insight. It sounds like a logical trade. But it depends on something we rarely talk about: whether people are actually developing the thinking skills that higher-value work requires.
McKinsey’s State of Organizations 2026 reports that when employees’ time is freed by AI, only 30% of leaders say they see that capacity being redirected toward higher-value activities such as critical thinking and creativity. It’s a useful reminder that creating capacity for higher-value work is not the same as ensuring that people are prepared to do that work well.
At the same time, AI is changing which parts of the work people do themselves. A 2025 study from Carnegie Mellon University and Microsoft Research found that GenAI shifted cognitive effort away from some aspects of task execution and instead toward verification, integration, and oversight. The same study found that greater confidence in AI was associated with less, not more, critical-thinking effort.
That creates a troubling tension. As AI takes on more routine work, the tasks that remain with people increasingly require judgment, interpretation, and critical thinking. Yet those are the very skills we may be exercising less as AI takes over more of the mental work that once helped us develop and use them.
That creates a practical question as we prepare our workforce for the future: If interpretation, synthesis, discernment, and sound decision-making are becoming more consequential, where are people getting enough opportunities to practice them?
Critical Thinking Is More Than Checking Whether AI Is Right
Much of the conversation about critical thinking and AI has focused on verification: Check the facts, watch for hallucinations, look for bias, and do not assume a fluent answer is a correct one. Those are necessary habits, but they describe only the final quality-control layer.
The harder thinking often happens earlier.
- It happens when someone has to decide what the problem actually is before choosing a solution.
- It happens when two explanations fit the same evidence, and neither can be dismissed quickly.
- It happens when information from different sources needs to be connected into a coherent view, or when a polished recommendation needs to be questioned because something important feels absent even though nothing is obviously false.
Those capabilities are difficult to build through explanation alone. Telling people to “challenge assumptions” is not the same as repeatedly putting them in situations where their assumptions are exposed. Telling people to “consider other perspectives” is not the same as having to reconcile a credible interpretation that conflicts with their own.
The research reviewed in the Carnegie Mellon/Microsoft study points to approaches such as structured argumentation, questioning, and discussion involving conflicting perspectives as ways to encourage critical and reflective thinking. That gives L&D a better starting point than another reminder to “think critically.” When learners’ work depends on judgment, interpretation, or synthesis, we can design experiences that give them more practice doing those things for themselves.
Here are five ways that can look in practice.
1. Give Learners Something Worth Interpreting
Many training scenarios do most of the sense-making for the learner. They explain what the problem is, then ask what should happen next. That works when the goal is to apply a clear procedure, but it’s less useful when the job requires people to diagnose an ambiguous situation before they act.
Consider a leadership program on a performance conversation. A typical scenario might tell learners that an employee has become disengaged and ask them to choose the best response. The learner is practicing how to respond to disengagement, but not how to determine whether disengagement is actually the problem.
A version requiring people to exercise more critical thinking would give learners the evidence instead: missed deadlines, a recent workload increase, changes in responsibilities, comments from colleagues, performance history, and what the employee has said. Then have learners answer: What do you think may be happening? Which evidence shaped your interpretation? What else would you want to know before deciding what to do?
The core leadership skill hasn’t changed. The learner still needs to decide how to handle the situation. But now they also have to interpret the situation rather than accept the diagnosis supplied by the course.
This is useful anywhere the job requires people to make sense of incomplete or conflicting information before acting, such as sales discovery, troubleshooting, customer escalations, investigations, coaching, consulting, or advisory work. In those cases, the opportunity for critical thinking is not an extra activity added to the training. It comes from giving learners enough of the problem to interpret for themselves.
2. Let Learners Encounter Another Person’s Reasoning
We often think of critical thinking as something that happens inside one person’s head. In practice, some of the best thinking happens when another person sees the situation differently. They notice something we missed, question an assumption we did not realize we were making, or interpret the same evidence through a different lens.
That gives peer discussion a more useful role than simply asking learners to “discuss the case.” Give two learners the same ambiguous situation and ask each to form an initial view on their own. Then have them compare how they got there. What did one person notice that the other did not? Where did their interpretations diverge? What assumptions were behind those differences? What new information would change either person’s mind?
For example, two learners might review the same customer complaint. One sees an unreasonable customer. The other sees a failure in the company’s process. The value is not in revealing the “right” answer as quickly as possible. It is in having each learner test their interpretation against another credible view and decide what the evidence actually supports.
This works especially well when the job itself involves ambiguity, competing stakeholder perspectives, negotiation, leadership judgment, or other situations where capable people can reasonably reach different conclusions.
3. Ask Learners to Make the Connection
Critical thinking is not only the ability to dismantle an argument. Some of our most valuable thinking comes from constructing connections between pieces of information that didn’t arrive already connected.
Training frequently does this cognitive work for the learner. We present several findings, explain what they mean collectively, and then ask a knowledge-check question about the conclusion. When synthesis is part of the role expectations, we can reverse that process and ask learners to build the connection themselves.
Give learners two findings, two cases, information from different parts of the business, or even examples from seemingly unrelated fields. Ask what becomes visible when they consider them together.
For example, imagine giving learners two observations: AI is reducing the time required for some administrative work, while relatively few leaders in a report see AI-freed time redirected toward critical thinking and creativity. Rather than explaining what those facts mean, ask learners to develop several hypotheses for why both could be true.
A more challenging version draws from another field. What might a customer-service operation learn from how aviation handles handoffs or exceptions? What might a project team learn from the way healthcare approaches error escalation? Learners then have to decide whether the analogy reveals a principle that is actually useful to transfer.
That means being able to explain why the connection holds, where it breaks down, and what new insight it offers.
4. Make Room for More Than One Plausible Explanation
AI is exceptionally good at turning a messy question into a coherent answer. Human beings like coherence too. The risk is reaching it too quickly.
As such, we can create learning experiences that help people become more comfortable keeping several possibilities open before selecting one. Instead of constructing every scenario around an obvious correct interpretation, we can present information that reasonably supports several hypotheses.
A regional sales decline, for example, might stem from weak execution, pricing, changing customer needs, a competitor move, or a product issue. Rather than immediately asking learners to select the cause, ask them to examine the possibilities. What evidence supports each explanation? What does each fail to explain? What information is still missing? What would they investigate next? What would cause them to change their current view?
The objective is not indecision. Eventually, people have to act. The capability being developed is knowing when the available evidence supports a conclusion and when the situation still requires more inquiry.
This type of activity is particularly useful in training for roles where ambiguity is part of performance. If the actual job always has a clear procedural answer, adding uncertainty simply makes the training unnecessarily difficult. But when employees genuinely need to diagnose, adapt, or weigh competing evidence, the learning experience should not always resolve that ambiguity for them.
5. Make the Reasoning Visible— & Give Learners a Chance to Revise It
One of the easiest ways to make a scenario more cognitively demanding is also one of the simplest: Don’t stop at the answer.
If a learner selects option B, we know very little about the quality of the thinking that produced it. They may have carefully considered the evidence, recognized a trade-off, and chosen B for a sound reason. They may also have guessed, remembered a phrase from the previous slide, or selected the response that sounded most professionally worded.
Where judgment matters, ask for a little more of the reasoning:
- What information influenced the decision most?
- What alternative did the learner seriously consider?
- What are they least certain about?
- What additional evidence would cause them to change their mind?
Then, where useful, change the situation. Introduce another piece of information, reveal an unexpected consequence, or provide feedback from an expert or peer. Allow learners to revise their conclusion.
The ability to revise is important because critical thinking is not demonstrated by stubbornly defending an original position. A learner who changes a decision when better evidence appears may be showing stronger judgment than someone who happened to select the correct answer immediately for a poor reason.
It also gives managers and learning teams something more meaningful to coach. Feedback can address not only whether the decision was appropriate but whether the learner noticed the relevant evidence, considered credible alternatives, and adjusted when the situation changed.
Look For the Moments Where Judgment Matters
None of these ideas requires creating a standalone “critical thinking.” In many cases, they’re relatively small changes to the learning already being designed.
- A scenario that currently provides the diagnosis can instead provide the evidence.
- A breakout discussion can ask learners to compare independently formed interpretations rather than simply “sharing thoughts.”
- A case study can ask learners to derive the connection instead of explaining it for them.
- A simulation can include several explanations instead of one clearly signposted path.
- A knowledge check can ask learners briefly why they made a choice and then introduce information that tests the reasoning behind it.
The common thread is not making learning harder. It’s looking at the performance itself and asking where good performance depends on thinking rather than simply following. It’s about using the situations that require learners to exercise discernment and not surrender their thinking. Those are the places where learners may benefit from more opportunities to interpret, compare, connect, question, and revise.
There are also many places where this is unnecessary. If the performance requirement is to follow a defined procedure, locate information, apply a clear policy, or complete a routine transaction accurately, the ambiguity in the activities can get in the way of learning. Critical-thinking practice is most useful when the work genuinely requires judgment, synthesis, diagnosis, adaptation, or decisions under uncertainty.
Higher-Value Work Still Has to Be Learned
The productivity promise of AI is appealing partly because it suggests that people will be able to spend less time on routine work and more time doing what humans do particularly well. But that future depends on more than giving people back hours in the day.
It depends on whether people can use that capacity to make sense of complexity, connect ideas, challenge an initial interpretation, and make sound judgments when there is no obvious answer. L&D can’t manufacture years of experience. We can, however, create more experiences in which people have to think: interpreting incomplete evidence, encountering another credible view, finding connections, staying with uncertainty long enough to investigate it, and revising a decision when new information warrants it.
If organizations increasingly need people to contribute judgment and critical thinking, those capabilities cannot remain aspirations attached to the AI productivity story. L&D can help ensure that as AI changes the work, people are still developing the judgment to do what matters most.
Image credit: Deagreez
