Human Judgment Can Be Trained. Here’s How To Build It.
Judgment is not something you either have or don’t
AI has made one thing clear that merely having a human review the output does not automatically improve it.
In Part 1 of the Judgment in an AI world series, we looked at why. People may not recognize that deeper scrutiny is needed. They may not have the time or reason to engage with the output properly. And sometimes they simply do not know enough about the subject to challenge what AI has produced.
That leaves us with a key question.
Can judgment be trained?
People are told to “use good judgment,” as if judgment were something that employees-even new hires- just have in store. Research suggests it is actually trainable.
Sir Andrew Likierman at London Business School has spent more than a decade studying workplace judgment, drawing on interviews with more than 800 people across more than 20 countries.
His framework breaks judgment into six components:
Knowledge and experience
Awareness of context
Trust
Feelings and beliefs
Choice
Delivery
When AI enters the decision, the framework can look like this:
AI can supply information, compare options, and produce analysis quickly. The harder parts still sit with the person using it. You have to understand the context around the answer. You have to decide how much trust it deserves. You have to notice the beliefs or biases affecting your own reading of it. You have to consider other options and judge whether the recommendation will actually work once it leaves the screen.
The point is not that every decision needs a six-step checklist. It is that judgment has parts and they can be practiced.
You can become better at checking your own beliefs.
You can become better at testing an answer against its context.
You can learn to notice when confidence is doing too much of the work.
You can practice looking for the missing option rather than choosing between the first two presented.
You can reflect on a decision subsequently and ask where your reasoning held up and where it did not.
How to train your mind to build better judgment
Judgment needs capacity before it needs technique. You can know how to challenge an assumption and still fail to do it when your attention is shot. You can understand the need to pause and still rush through the answer because five other things are waiting. You can know that context matters and still accept the first plausible recommendation because your mind is already working at its limit.
That is why judgment training cannot begin with a checklist of better questions. The first requirement is enough internal capacity to use them. We explored this more deeply in our piece on how AI use affects cognitive capacity, where we look at what happens when too much of the reasoning gets handed over to the tool.
Having internal capacity means being able to slow the automatic response, hold more than one interpretation in mind, notice when your own beliefs are shaping the answer, and stay with a difficult decision long enough to examine it properly. Those are the conditions that make the practical skills below usable.
Critical Thinking: challenge the first answer
One of the easiest ways to weaken judgment is to start with a conclusion and spend the rest of the process defending it.
That problem existed long before AI. But AI makes it worse because it gives you a polished starting position almost instantly.
This is where critical thinking- the skill to examine the evidence and test whether the reasoning comes into practice.
It can look like this:
Ask, “What would have to be true for this answer to hold?”
Then ask, “What would make the opposite conclusion reasonable?”
That second question forces you to look for evidence your first interpretation may have ignored.
Here are some practical ways to train your mind for critical thinking.
Name the assumption underneath.
Research suggests we don’t weigh evidence neutrally- people on opposite sides of an issue can read the very same mixed evidence and each walk away more convinced they were right (Lord, Ross & Lepper, 1979). Prior beliefs influence how we interpret what we see.
So make the assumption visible.
If AI recommends delaying a launch, what is that recommendation assuming about customer demand?
If it recommends hiring for a certain skill set, what is it assuming about how the role will develop?
If it says one strategy is strongest, what criteria has it treated as most important?
Another useful countermeasure is to deliberately argue against the answer you currently favor.
Researchers have tested this through a technique often called “consider the opposite.” For example, in an instance of classic work on judgment bias, simply telling people to be fair had little effect. Asking them to actively consider why their preferred conclusion might be wrong produced a stronger corrective effect.
Check the source, not just the claim.
Ask where the information came from, how reliable that source is, who benefits, and what may be missing before you use it. This is more urgent with generative AI than with a search engine, because the output reads as authoritative while showing you none of its sourcing.
A claim may be accurate, outdated, taken out of context, or based on a weak source that you would never have relied on if you had seen it yourself.
For anything that could affect a real decision, go back to the original source where possible. Check whether the evidence actually supports the way the AI has interpreted it, and whether there is context that changes the conclusion.
Do your own thinking first, then bring in the tool.
There is another useful habit here: do some thinking before asking AI what it thinks.
Form an initial view first. Even a rough one. Then use AI to challenge it, find gaps, generate alternatives, or test the reasoning.
That changes your relationship with the tool. You arrive with something to compare against instead of allowing its first answer to become the frame for everything that follows.
Research on cognitive offloading suggests that when more of the thinking is routinely handed over to an external tool, people can become less engaged in the reasoning itself. Deliberate use of tools gives you a better chance of keeping that reasoning active.
Put a pause between the answer and the decision
AI removes a lot of the waiting required for an output to get completed.
That is useful for speed. It also means generation, review, and action can start collapsing into one continuous motion.
Good judgment sometimes needs a break in that sequence.
Building the pause is the ability to stay with what is happening in the moment long enough to notice what deserves attention before acting.
In AI-assisted work, that can mean adding deliberate friction around decisions that carry higher consequences.
Before accepting the output, ask:
- What assumption is this making?
- What evidence would make this wrong?
- What context is missing?
- What happens if this recommendation fails?
- Am I agreeing because the reasoning holds, or because the answer sounds convincing?
The aim is not to turn every AI output into a formal review. The level of friction should match the stakes.
A rough internal brainstorm may need a few minutes. A client recommendation, hiring decision, financial forecast, or strategic call deserves more detailed scrutiny.
The skill is knowing when to stop treating speed as the goal.
Review the reasoning after the decision
Judgment improves when you look back at how you reached the decision, rather than only whether the result turned out well.
A good result can come from weak reasoning. A poor result can come from sound reasoning in a situation where the outcome was uncertain.
If you only focus on the end result, you miss this useful part of how you got to the result.
After an important call, ask:
What did I expect to happen?
What actually happened?
Which part of my reasoning held up?
Where did I make an assumption that turned out to be wrong?
What will I check earlier next time?
A short decision journal or after-action review can make this much more concrete and be your tool to keep building on the ‘skill of good judgment’.
This is also one of the best ways to make judgment visible inside an organization.
KPMG recommends having employees record whether they accepted, modified, or rejected an AI output, along with the reasoning and criteria behind the choice. That gives managers something they can actually discuss and coach.
Instead of saying, “use more judgment,” a manager can ask why a recommendation was accepted, which assumption was challenged, or what evidence caused the employee to override the AI.
That turns an invisible cognitive process into something people can examine and improve.
Judgment needs practice in real work
Judgment gets stronger through repetition. The logic is that people need repeated chances to apply judgment where the decision has context, pressure, and consequences.
That is also where organizations need to change how they think about AI training.
Tool fluency still matters, but people also need practice with the skills that determine how the answer is reached and what happens the answer appears:
- Applying critical thinking to the process and output and challenging its validity, relevance, and accuracy
- Creating enough space, mental and physical, for an adequate level of scrutiny
- Reviewing your own reasoning, after the fact, and improving the ‘judgment process’
The capacity to make better judgment becomes stronger when these techniques are used inside real decisions, followed by a feedback loop.
Judgment does not improve because someone was told to “be more critical.”
It improves when the thinking behind a decision becomes something people can practice, inspect, and refine.
Conclusion: Judgment is becoming part of the job
AI can produce more of the work before a decision. That makes the quality of the human thinking around that work way more crucial.
A few things are worth carrying forward:
- Judgment needs scrutiny. A plausible answer still needs someone willing to test it.
- Judgment depends on knowledge. You cannot challenge an output well if you do not understand the work behind it.
- Judgment has trainable parts. Concrete ways to question, pause, reframe, and review their reasoning.
- Practice has to happen in real work. Judgment improves when people apply these skills to real decisions and get feedback on how they thought through them.
This is where Q Studio’s Mind Skills Training™ fits in.
Team Adaptability helps teams build the Mind Skills needed to think and respond well as AI changes the way work gets done. Leadership Capacity helps leaders strengthen their capacity – a prerequisite for sound judgment, especially in higher-stakes decisions.
If AI is taking over more of the production inside your organization, the human investment has to keep pace. Q Studio works with leaders and professionals to build the Mind Skills behind better judgment, so people know what to question, what to trust, and what deserves to become action.