Walk the floor at ATD26 and count the booths selling leadership training versus everything else. I did. It wasn’t close. That’s not a coincidence; it’s a signal. AI is eating the routine work in L&D, which means judgment, the ability to tell a good answer from a confident-sounding bad one, just became the whole game. I spent my session, “AI in L&D: Tools and Strategies You Can Use Now,” showing the room tools that build content and coach people faster than any of us could a year ago. What’s stuck with me since is a simpler question: what happens to the person who never had to build that judgment in the first place, because the tool did the thinking for them?
Why Does Critical Thinking Matter More as AI Takes Over Routine L&D Work?
Here’s the mechanism: AI removes the routine reps that used to build judgment on the job. A model can draft a report or a lesson in seconds. The differentiator isn’t who can generate that draft anymore; it’s who can look at it and tell whether it’s right, half-right, or confidently wrong. That’s not a skill you pick up by accident.
I counted the exhibitors on the ATD26 floor selling some version of leadership or critical-thinking training. It was most of them. That tells you where the industry thinks the gap is actually opening, and it’s not tool access. Organizations already have plenty of AI tools. What they don’t have enough of, in my experience, is people equipped to catch a bad answer before it goes out the door. And that shortage won’t show up on a training completion dashboard until it’s already cost you something.
AI Insight: If your L&D metrics are all about content velocity, you might be measuring the wrong side of this. Judgment is the thing that quietly erodes while output speeds up.
The Bottleneck Isn’t the AI Tool, It’s Your Data
I’ve started thinking of AI in L&D less as a technology question and more as a mirror. Most organizations can tell you someone’s job title in the LMS and which courses they’ve finished. What they can’t tell you, in that same system, is how that person is actually performing, what their manager thinks of them, or what they’re capable of that hasn’t shown up in a completion record yet.
Hand a model a thin picture and it’s not being lazy when it gives you a generic answer back. It’s doing the best it can with what you gave it. Which is exactly why the judgment argument matters so much right now: if your data can’t give AI enough context to be trusted on its own, the human checking its work has to be good at catching what it missed. That’s a capability you build on purpose. You don’t get it by accident because someone happened to be in the room.
How Do You Use a Tool Like Claude Without Losing the Judgment You’re Trying to Build?
I use Claude the way I’d brief a new instructional designer. Give it the topic, the audience, the time limit, then keep talking to it like a person. The judgment stays intact as long as the process ends with a subject matter expert reviewing the draft, not with the model’s output going straight to learners unchecked.
On stage at ATD26, I started from a deliberately thin, two-sentence prompt, just to show the room what minimal context actually produces. Then I kept building: a short video for a term the audience wouldn’t recognize, a quiz, an easier version of that quiz, an adjustment to fit our sales methodology. None of that took technical skill. It took being specific and a little curious. The step I care about most came last: packaging the draft for a subject matter expert to react to. That review isn’t just quality control, it’s the rep that builds the SME’s own judgment for next time. There’s no such thing as AI plus AI yet. At least not one I’d trust unsupervised in a room full of people trying to learn something that matters.
Content Tool or Coaching Tool? Why That’s a Talent Strategy Decision, Not a Tech One
Not every AI use case in L&D is the same shape, and I’ve started treating that distinction as a talent strategy call, not a technical one. A tool built to answer a question should just answer it. A tool built to develop a person should do the opposite: lead them toward the answer instead of handing it over.
I tested this live with a coaching agent responding to a scenario about a chronically late employee. A respectful opening got a calm, guided response. A deliberately rude opening (on purpose, for the demo) got a visibly shaken reaction that showed the room exactly what a bad opening does to a coaching moment.
If your organization is trying to build bench strength and not just answer questions faster, that’s where your AI investment actually has to go. At Infopro Learning, this is the thinking behind our Intelligent Design Framework: a coaching agent loaded with your actual leadership model, built to develop people rather than just inform them.
AI Insight: The most common mistake I see isn’t picking the wrong AI tool. It’s asking one tool to do two fundamentally different jobs: one that answers, and one that’s supposed to develop someone.
Is It Safe to Use Free AI Tools Like NotebookLM for Enterprise Training?
NotebookLM is free and, unusually, it doesn’t train on the data you feed it. That’s what makes me comfortable recommending it for personal experimentation. It is not, however, enterprise clearance. Paid tools generally keep your data inside your own environment, but IT still has to sign off before real learner or client data goes anywhere near it.
My rule of thumb with free tools: if it’s free, you’re usually paying with your data. Where NotebookLM has genuinely impressed me is in personalizing how training gets consumed. Feed it a finished module and, in a couple of clicks, it’ll generate a podcast version, an infographic, and a short video from the same material. Not everyone learns the same way, and getting that flexibility for free, in the time it takes to click a button, is one of the more useful things I’ve watched AI do in this space.
Where I’d Start, If Retention Is the Goal
- Pick one real build, not a demo. Give the tool a real brief: topic, audience, time constraint, any methodology or compliance requirement it has to fit. A vague prompt gets a vague answer, and vague answers don’t build anyone’s judgment.
- Decide what you’re developing before you pick the tool. Content and coaching need different designs. If bench strength is the actual goal, that decision comes before the shopping, not after.
- Get one enterprise-tier tool formally cleared. Free tools are fine to learn on. Real learner and client data needs something your organization has actually approved.
I’d aim to cut a ten-day build down to five, not down to one.Spend the time you save building the judgment your team will need once AI has taken over everything routine, because that’s the part nobody’s LMS dashboard is tracking yet.
Frequently Asked Questions (FAQs)
-
remove Why is critical thinking becoming more important in the age of AI?AI can quickly generate answers, summarize information, and automate routine tasks, but it cannot consistently apply human judgment, ethical reasoning, or contextual decision-making. Critical thinking enables employees to evaluate AI-generated outputs, identify inaccuracies, and make informed business decisions. Organizations that strengthen critical thinking skills are better equipped to improve innovation, reduce risks, and build a workforce that adapts to rapid technological change.
-
add How does critical thinking support employee retention?Employees are more likely to stay with organizations that invest in meaningful skill development rather than relying solely on automation. Critical thinking empowers individuals to solve complex problems, contribute strategic ideas, and grow into leadership roles. This creates stronger career development opportunities, higher job satisfaction, and increased engagement—key factors that improve employee retention.
-
add Can AI replace critical thinking in the workplace?No. While AI excels at processing data and generating recommendations, it lacks human qualities such as reasoning, creativity, empathy, and ethical judgment. Critical thinking remains essential for interpreting AI insights, challenging assumptions, and making decisions aligned with business objectives. The most successful organizations combine AI capabilities with human critical thinking to achieve better outcomes.
About The Author
Nolan Hout
Senior Vice President, Growth & AI Strategist at Infopro Learning
Nolan Hout has over a decade of experience in the L&D industry, helping global organizations to unlock the potential of their workforce. Nolan is results-driven, investing most of his time in finding ways to identify and improve the performance of learning programs through the lens of return-on-investment. A former Forbes Council Member, he is passionate about networking with people in the learning and training community and he grows this network through a popular podcast called ‘The Talent Equation with Nolan Hout’, where he interviews executives about any topic related to talent development. On a personal note, Nolan is an avid outdoorsman and fly fisherman, spending most of his free time on rivers across the Pacific Northwest.
