Adults Learn Through Problems, Not Topics

Start with the problem, then teach the model.

Most training is organized around topics.

Communication. Leadership. Feedback. Conflict. Time management. Customer service. Decision-making.

That makes sense from the trainer’s side. Topics are easy to organize. They look clean in an agenda. They fit neatly into modules. They make the designer feel like the session has structure.

But adults do not usually wake up thinking, “I need a deeper understanding of communication theory today.”

They think, “I need to tell my employee they missed the standard without starting a fight.”

They think, “My team is stuck and I do not know how to get them moving.”

They think, “I have a customer who is angry, and I need to keep this from getting worse.”

They think, “I keep having the same meeting over and over, and nothing changes.”

Adults are not usually looking for a topic.

They are looking for help with a problem.

That is why problem-centered learning matters. It moves training closer to the way adults actually experience work. They do not live inside neat categories. They live inside messy situations that require judgment, choices, tradeoffs, and action.

AI can help us design for that.

It can also make the problem worse.

Topics are clean. Work is not.

There is nothing wrong with teaching topics.

The problem comes when the topic becomes the center of the learning experience instead of the work people need to do.

A content-centered session usually starts with the subject. The trainer explains the concept, gives background, introduces a model, walks through the main points, then maybe gets to application near the end.

That feels logical.

It is also often backwards.

A problem-centered session starts with the real situation. It gives learners something to wrestle with early. Then the model, concept, or framework becomes useful because it helps them handle the problem better.

That is a different experience.

In a content-centered session, the learner is often asking, “Why are we talking about this?”

In a problem-centered session, the learner is asking, “What would I do here?”

That second question is far more valuable.

It creates attention. It exposes assumptions. It makes people reach for experience. It reveals gaps. It gives the trainer something real to coach.

This does not mean we abandon content. It means content serves the problem, not the other way around.

Teaching a topic is not the same as helping solve a problem

Here is a simple example.

A trainer teaching “feedback” might explain a feedback model, define effective feedback, show a few examples, and ask participants to discuss why feedback matters.

That may be fine.

But the real problem might be this:

A new manager has an employee who is friendly, well-liked, and consistently missing deadlines. The manager has already hinted at the issue twice, but nothing changed. Now resentment is building on the team, and the manager does not want to damage the relationship.

That is not “feedback” as a topic.

That is a problem.

It has pressure. It has history. It has risk. It has emotional tension. It has consequences if the leader avoids it.

Now the feedback model has a job to do.

The learner is not memorizing a concept. They are using the concept to make a better move.

That is where adult learning gets stronger.

The same thing is true with almost any topic.

Conflict becomes: “Two high performers are undermining each other in meetings.”

Delegation becomes: “You keep taking work back because your team member does it slower than you.”

Communication becomes: “Your department keeps blaming another department, and both sides think they are right.”

Decision-making becomes: “You have incomplete information, two bad options, and a deadline by noon.”

Those are not topics.

Those are training moments.

Where AI actually helps

AI is very useful for building problem-centered training, especially when you give it real constraints.

It can help create cases, dilemmas, branching choices, troubleshooting exercises, and realistic practice situations.

That matters because building good scenarios takes time.

A trainer might know the topic, but struggle to create enough good practice material. AI can help by generating several versions quickly. Beginner version. Advanced version. Frontline version. Executive version. High-pressure version. Low-trust team version.

That kind of variation matters because adults need problems that feel close enough to their world.

AI can also help create branching choices.

For example, you can give learners a situation and ask them to choose between three possible responses. Each choice leads to a consequence. Then learners have to adjust.

That is much better than asking, “What should a good leader do?”

A good problem forces a decision.

AI can also help you add friction.

You can ask it to make the scenario messier. Add time pressure. Add competing priorities. Add a reluctant employee. Add an unclear policy. Add a senior leader who wants a quick answer. Add a customer who is upset but partly right.

That is where the scenario starts to feel more like work.

The goal is not drama. The goal is realism.

Adults do not need perfect stories. They need practice with the kind of imperfection they face every day.

Where AI makes this worse

AI-generated scenarios often have one big flaw.

They are too clean.

The problem is obvious. The bad behavior is obvious. The right answer is obvious. The language is polished. The participants sound like characters in a corporate training video.

That kind of scenario does not stretch judgment.

It only lets learners perform agreement.

Everyone knows what they are supposed to say. The group gets through the exercise. The trainer checks the box. Nothing difficult happened.

That is fake problem-centered learning.

The problem looked like a problem, but it did not require real thinking.

AI can also make scenarios too generic.

A “manager struggling with communication” is not enough. A “team experiencing resistance to change” is not enough. A “customer upset about service” is not enough.

Those are placeholders.

Real problems have texture.

Who is involved. What happened before. What is at stake. What makes this difficult. What would a reasonable person be tempted to do wrong. What information is missing. What pressure is shaping the decision.

If those pieces are missing, the activity will feel artificial.

And adults can tell.

They may still participate, but they will not fully invest. They know when a case has been built from real work and when it has been assembled from training clichés.

The standard for a useful training problem

A strong training problem usually has a few features.

First, it has a recognizable situation.

Learners should be able to say, “I have seen something like this,” even if the details are not identical.

Second, it requires a decision.

If the scenario only asks people to talk about principles, it is probably too soft. Ask them what they would do next. What they would say. What they would write. What they would change. What they would stop doing.

Third, it includes constraints.

Time, people, policies, politics, incomplete information, limited authority, competing priorities. Constraints are what make judgment necessary.

Fourth, it has more than one tempting wrong answer.

That is important. If the bad option is cartoonishly bad, the exercise becomes too easy. Good training problems include mistakes that reasonable people might actually make.

Fifth, it leads naturally into the concept you are teaching.

The problem should create the need for the content. Learners should feel, “We need a better way to think about this.” Then the model earns its place.

That is when content lands.

What this looks like in practice

Suppose you are teaching conflict management.

The content-centered version starts with definitions, types of conflict, communication styles, and maybe a model for resolution.

Again, not useless.

But it may take too long to matter.

The problem-centered version starts with a case.

Two team leads are disagreeing publicly in meetings. One says the other is careless with deadlines. The other says the deadlines keep changing without warning. Their teams are starting to take sides. A director has asked you to “get them aligned” before Friday’s planning meeting.

Now ask the learners:

What is the first conversation you would have?

What information do you need before you intervene?

What would make this worse?

What standard needs to be clarified?

What would you say if one person tries to pull you into blaming the other?

Now the conflict model matters.

Learners are not studying conflict in general. They are trying to handle this conflict with better judgment.

AI can help you build three variations of that same case. One for new supervisors. One for senior leaders. One for cross-functional teams.

But you still have to choose, edit, and sharpen the case.

AI can draft it.

The trainer has to make it true.

Use it this week

Take one topic you are scheduled to teach.

Before you write another slide, ask this:

What problem does this topic help someone solve?

Write the problem in plain language.

Then use AI to generate three workplace scenarios based on that problem. Do not accept the first version. Push it.

Ask AI to add constraints.

Ask it to make the first obvious solution create a new problem.

Ask it to include incomplete information.

Ask it to create a realistic consequence if the learner handles it poorly.

Then pick the best scenario and put it early in the session.

Let learners wrestle with it before you explain everything.

That may feel uncomfortable if you are used to teaching the topic first. Good. That discomfort is often where better learning begins.

The trainer still matters

AI can produce cases quickly.

That does not mean they are good.

You still have to ask whether the case fits your learners, your objective, and the real world they operate in. You still have to remove the corporate training smell. You still have to make sure the scenario requires the kind of thinking and behavior you want to develop.

And you still have to facilitate the debrief.

That is where the learning often happens.

Not in the case itself, but in the discussion after the decision.

Why did you choose that move?

What were you assuming?

What would happen if you tried that at work?

What risk did you miss?

What would a more experienced person notice?

How does the model help now?

Those questions turn the activity from entertainment into learning.

The standard

If you remember one thing from this chapter, make it this:

Adults do not need more topics. They need better problems to solve.

AI can help you build those problems faster.

It can give you scenarios, dilemmas, branching decisions, and practice cases. It can help you adapt them for different roles and experience levels.

But AI will also hand you clean, generic, lifeless problems if you let it.

Do not let it.

Make the problem real enough to respect the learner.

Make the decision hard enough to require judgment.

Make the content useful enough to earn its place.

That is problem-centered learning.

Not a topic with an activity attached.

A real problem that makes the learning necessary.