Critical thinking and AI: 7 habits for better decision-making

Artificial intelligence can write a report, analyze data or suggest an action plan in seconds. And it often does so with impressive confidence. The problem? Just because an answer sounds intelligent doesn’t mean it’s one we should rely on to make decisions.

That’s where critical thinking comes in. Yet developing this skill doesn’t always seem to be keeping pace with AI adoption.

According to a 2026 Boostalab report, 23% of respondents consider critical thinking a priority skill to develop among non-management employees, but only 5% perceive their organizations as investing in its development (1). This 18-percentage-point gap, based on survey respondents’ perceptions, is worth a closer look.

After all, AI may help us save time, but we still need to make sure we’re not just making bad decisions faster!

What is critical thinking in the age of AI?

Critical thinking is the ability to take a step back, examine information, assess its credibility, question arguments and draw informed conclusions. It calls for analytical skills, curiosity, openness to different perspectives and an awareness of our own biases.

At work, critical thinking comes into play in all kinds of situations: interpreting survey results, solving complex problems, evaluating recommendations or allocating limited resources.

Developing critical thinking skills doesn’t mean questioning everything. It means learning to distinguish reliable information from unreliable information and recognizing what deserves a closer look before taking action.

Generative AI makes this skill particularly important. In the past, analyzing a situation generally required some research and reflection. Now, an initial recommendation can appear almost instantly.

That speed is valuable. But it also means we need to evaluate content we haven’t necessarily developed ourselves.

Why can a convincing AI response be misleading?

Imagine you’ve noticed a drop in employee engagement on your team. You ask an AI tool to suggest ways to address it.

A few seconds later, you have a polished action plan: one-on-one meetings, employee recognition activities and team-building initiatives. It’s enough to make you want to put the plan into action right away!

But here’s the catch: AI may not know that your team is already overloaded and that several employees have been asking for more autonomy. Is the response wrong? Not necessarily. But is it right for your situation? That’s another story.

Errors aren’t always obvious

Generative AI can confidently produce incorrect information. This phenomenon is often referred to as an AI “hallucination.” The National Institute of Standards and Technology has documented this phenomenon, along with other reliability concerns associated with generative AI (2). Several issues can find their way into a response:

• Inaccurate information: Incorrect statistics, fabricated references or claims that can’t be verified.
• Bias: Certain perspectives being favoured over others or questionable generalizations.
• Omissions: Important factors being left out of the analysis.
• Oversimplification: Solutions being presented as universally applicable when the reality is more complex.

Even when the facts are accurate, the reasoning can still be flawed. For example, observing that highly engaged teams hold more one-on-one meetings doesn’t prove that those meetings are what drive their engagement.

Fact-checking is essential. But critical thinking also means asking whether the information actually supports the conclusion being drawn.

How can you think critically when using AI? 7 habits to adopt

You don’t need to be an artificial intelligence expert to think critically at work. A few simple habits can make all the difference.

1. Clarify the real problem before looking for an answer

Before turning to AI, ask yourself: “What problem am I really trying to solve?”

Let’s go back to our employee engagement example. If you ask AI how to motivate a team, it will probably suggest motivation strategies. Makes sense! But if the real issue is an ongoing excessive workload, those solutions could miss the point entirely.

Start by separating symptoms from possible causes. What do you know for sure? Which assumptions still need to be confirmed? You can then use AI to explore different explanations instead of jumping straight to a solution.

The habit to build: Don’t let the speed of an answer distract you from the importance of asking the right question.

2. Check the sources instead of taking AI at its word

AI claims that a particular practice boosts productivity by 30%? That a study proves the benefits of a certain method? That a legal requirement applies to your organization? Before relying on that information, verify it.

And keep in mind that asking AI “Are you sure?” isn’t independent verification. The tool might simply rephrase its original claim with just as much confidence.

Go straight to the original sources whenever possible: scientific studies, publications from reputable institutions or official data. Check when they were published, the context they cover and any limitations.

The habit to build: The more a claim could influence an important decision, the more carefully it needs to be verified.

3. Spot hidden assumptions

AI responses often rely on assumptions that aren’t explicitly stated. A recommendation to automate a process might assume that the tasks involved are repetitive. A restructuring proposal might take for granted that everyone’s roles are clearly defined. But those conditions aren’t always in place. To uncover these assumptions, ask yourself: “What would need to be true for this recommendation to work?”

In the case of our disengaged team, suggesting more meetings assumes that employees have enough time for them and that a lack of communication is actually the problem.

The habit to build: Make sure the conditions behind a recommendation are realistic before weighing its benefits.

4. Look for missing perspectives

AI can present a convincing picture while overlooking certain points of view.

Imagine it recommends reducing remote work to improve collaboration. Has it considered the impact on concentration, differences between job roles or employees’ individual needs?

This habit is especially important when an answer confirms exactly what you already believe. That’s confirmation bias: our tendency to favour information that reinforces our existing beliefs. To counter it, make a point of looking for opposing arguments or alternative interpretations.

The habit to build: Be wary of responses that simply tell you what you want to hear without offering a more nuanced perspective.

5. Examine the reasoning, not just the conclusion

A clear, well-structured recommendation isn’t necessarily based on sound reasoning.

Consider this statement: “Companies that invest more in training perform better. Therefore, we should increase our training budget to improve results.”

The first claim might be supported by data. But the second needs a closer look. What types of training were studied? Could other factors explain the results? Does the proposed investment actually address the organization’s needs?

The habit to build: Look at how the reasoning connects observations to conclusions. Correlation doesn’t automatically mean causation.

6. Check the response against your workplace reality

AI can draw on a wealth of general information, but it doesn’t automatically understand the ins and outs of your organization. A change management initiative might look great on paper but be unrealistic for a team already dealing with several changes. A recognition strategy might be appreciated in one department but poorly received in another.

That’s where professional experience, listening to others and understanding the context become essential. Talk to the people involved, assess the constraints and consider the possible consequences.

The habit to build: A good solution isn’t just one that works in theory. It’s one that makes sense for your situation.

7. Take responsibility for the decision

Here’s perhaps the most important question: “Would I feel comfortable defending this recommendation to the people affected by it?”

Decisions about hiring, performance evaluations or working conditions deserve special attention. AI can help organize information, compare options or prepare questions. But sensitive decisions require thorough human analysis and, depending on the situation, input from specialists.

The habit to build: Treat AI suggestions as input to evaluate, not as a way to hand over responsibility for a decision.

Checklist for evaluating AI-generated content

Before using an AI-generated response at work, ask yourself these seven questions:

This checklist is a handy reference, not a guarantee of reliability. The more serious the potential consequences, the more thorough the verification should be.

Is AI weakening our ability to think?

We already rely on calendars to remember things, software to do calculations and navigation apps to find our way around.

This kind of delegation, known as cognitive offloading, can free up time and mental energy. But what happens when we start delegating the very activities that help us develop our reasoning skills?

The risk of skipping essential steps in the learning process

In 2025, a study conducted by Microsoft Research and Carnegie Mellon University examined the relationship between generative AI use and critical thinking at work (3). Based on 936 examples of AI use collected from 319 participants, the researchers found that the more confidence participants reported having in AI, the less they reported using critical thinking when completing those tasks.

The study also revealed a shift in mental effort. Rather than developing an entire analysis themselves, participants spent more time verifying and integrating AI-generated responses.

These findings are based on self-reported data and don’t prove that AI causes a lasting decline in cognitive abilities. They do, however, point to a potential risk: losing opportunities to practise our reasoning skills.

On the other hand, an experimental study published in 2026 by the Institute of Labor Economics suggests that AI can support learning when it helps people understand concepts rather than simply doing the work for them (4).

That distinction matters. Using AI to explore a question or challenge your assumptions can deepen your thinking. But consistently letting it do the initial analysis may limit opportunities to learn. So the issue isn’t just how often we use AI, but how we use it.

How can you develop critical thinking skills at work?

Thinking critically doesn’t mean turning every task into a full-blown philosophical exercise! A few simple habits can help make more room for reflection.

Three simple exercises to try

1. Think for two minutes before turning to AI
Before asking AI for a recommendation, come up with your own hypothesis or identify what information is missing. Then compare your thinking with its response. The goal is to stay actively involved in the analysis.

2. Play devil’s advocate
Ask AI to come up with three strong arguments against a proposal you think is excellent. Then consider those objections. Do they reveal risks you hadn’t thought of? Do they change your conclusion?

3. Compare perspectives as a team
Share an AI-generated recommendation with a few colleagues. Ask everyone to evaluate it individually, then compare your observations. The discussion may uncover blind spots that no one would have spotted on their own.

Make critical thinking a shared responsibility

These exercises are useful, but developing critical thinking skills over the long term also requires the right conditions. And organizations have an important role to play!

As AI speeds up content creation, setting clear expectations for quality and verification becomes even more important. Who is responsible for checking the information? When is a more thorough review needed? Do teams have enough time to use their judgment?

After all, asking teams to think critically while rewarding them only for working faster sends a pretty mixed message! Organizations therefore benefit from encouraging questions, creating space for respectful debate and giving people opportunities to practise critical thinking.

The goal: make critical thinking a shared habit rather than leaving it up to each individual.

With AI, better decisions start with better questions

Artificial intelligence helps us move faster. But that doesn’t mean we can stop thinking about where we’re headed. The seven habits covered here remind us of one essential thing: the quality of a decision isn’t measured by how quickly we get an answer.

Developing critical thinking skills helps maintain the quality of our work, preserve learning opportunities and strengthen professional judgment. And this skill deserves more than a simple reminder to “be careful.”

Critical thinking develops through practice, discussion and support. That’s exactly what Boostalab’s Developing critical thinking in the age of AI training is designed to do: help teams sharpen their judgment and use AI without handing over their decision-making responsibilities.

Because at the end of the day, AI can make suggestions. But the decisions are still ours to make.

References

1. Boostalab. (2026). Compétences humaines 2030 : ce que les organisations québécoises doivent développer à l’ère de l’IA. Boostalab.
2. National Institute of Standards and Technology. (2024). Artificial intelligence risk management framework: Generative artificial intelligence profile. NIST.
3. Lee, H.-P., et al. (2025). The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. 2025 CHI Conference on Human Factors in Computing Systems.
4. Contractor, Z., & Reyes, G. (2026). Experimental evidence on the learning impact of generative AI. Institute of Labor Economics.