BALANCING AI ADVANCEMENT Vs DEVELOPMENT

What We Stop Practicing

I was standing in the checkout line a few months ago when the register froze.

The cashier poked the screen like it was a whack-a-mole trying to get jolt it back to life as it dawned on her that she might have to check us all out manually.

Normally, she’d direct us to the self-checkout line but it was closed because it was costing the company more in theft than it was saving in wages. So screen stuck, line multiplying, impatience bubbling, she turned to the cash in hand and tried to figure out the change.

Fortunately, the system froze after it had calculated the amount of change, but it hadn’t yet specified the exact number of each bill and coin required. You could see her trying to work through it, hesitating, second guessing.

Not because she wasn’t capable, but because she didn’t need to know how. The system had been doing that part for her. Most of the time, it works! It’s faster, more consistent, and has fewer mistakes.

Until it stops working.

And as the line multiplied, I saw something clearly: what we’ve stopped practicing.

We’ve been heading in this direction for decades.

Calculators, GPS, systems that anticipate the next step and take care of it for us. None of it felt like a huge deal on its own. It made things easier, more efficient.

It’s not that people stop caring or trying (although many of us did). It’s that certain skills just don’t get used anymore. And when that happens long enough, they start to go.

I heard a comedian say recently—half joking, half warning—that we should all try to stay healthy because our future doctors are currently using AI to get through their exams.

I laughed as my gut sank. It tapped to something people are starting to feel. Not just about their own work, but about everyone else’s.

If AI is helping everyone do their job, where is the actual capability being built?

And maybe more importantly—how much of it do we really trust?

There’s a scene in Westworld that’s stayed with me for a similar reason.

A futuristic ambulance shows up and everything about it is advanced. The technology can diagnose, guide treatment, move things along quickly. The EMTs are there, but they don’t even know how to check for a pulse. They’re the drones following AI’s instructions.

It all works seamlessly.

Until something unexpected happens. Or a system goes down. Or technology costs spike.

And then it’s not obvious who actually knows what to do.

The scene feels off in a way that’s hard to explain. Not because it’s unrealistic. Because it feels just close enough to plausible.

AI isn’t creating something entirely new here—it’s accelerating something that was already happening.

Only now it’s not just tasks or calculations and it’s not taking directions. It’s directing us and we trust it.

Not just helping us do things faster, but reshaping what we need to know in order to do them at all.

And that’s where the question starts to change.

Not “what can AI do for us?”

But “what happens to us when we rely on it to do more of the thinking?”

That question matters a lot in talent development.

Because development was never just about getting to the answer. It was about how you got there—the pattern recognition, the judgment, the ability to work through something when it isn’t obvious.

Learning to write before you’re good at it.

To diagnose before you’re right.

To make a call when you don’t have the full picture.

If AI is stepping in earlier—structuring the answer, refining the thinking, sometimes even providing the reasoning—then what part of that process is the person actually practicing?

And if they’re not practicing it, do they build it?

There are some early signals that this dynamic is already showing up in subtle ways.

When people rely heavily on AI to do their work, many report feeling like the system is doing most of the thinking—and their confidence in their own reasoning starts to slip.

That part is easy to overlook.

Because the output still looks good. Sometimes even better than before.

But underneath that, something is shifting.

People don’t feel as certain about how they got there.

And when that confidence starts to fade, we tend to respond in a pretty human way.

We lean on the system more.

Not because we’re suddenly less capable, but because we’re less sure. And the more we rely on it, the less we use our own judgment. The less we use it, the harder it feels to access.

It doesn’t happen all at once. It just… tightens over time.

Until eventually the system isn’t just supporting the work—it’s carrying the part we no longer trust ourselves to do.

One way to counter that drift might feel surprisingly familiar.

In the scientific method, we don’t just form a hypothesis and look for evidence that supports it. We actively try to prove ourselves wrong. We test, we challenge, we look for where something breaks.

That discipline matters because it keeps our thinking sharp. It forces us to stay engaged in the process, not just accept the outcome.

There’s a similar opportunity in how we use AI.

Instead of accepting outputs at face value, we can treat them as a starting point.

Something to question, stress-test, and refine. A hypothesis to work against, not just an answer to move forward with.

It’s a small shift in behavior, but it matters.
Because it keeps us in the loop of thinking. It strengthens our own reasoning rather than slowly replacing it.

And over time, it may be one of the ways we avoid turning assistance into dependence.

At the same time, it’s not just changing how people learn. In some cases, it’s changing whether they get the chance to learn at all.

A lot of the work that used to sit at the entry level—the repetitive, foundational stuff—is exactly what AI is best at. Which makes it tempting to automate it or shift it upward instead of staffing it.

On paper, that makes sense. It’s efficient. It reduces cost. It speeds things up.

But that entry-level work was never just about getting tasks done. It was where people learned how things actually work. Where they built judgment, context, intuition.

When that layer thins out, you don’t just save money today. You lose the foundation for who becomes experienced later.

And that gap doesn’t show up immediately. It shows up a few years down the line, when the next level up doesn’t have the depth you expect it to.

You can see a version of this in how work is changing in places like contact centers.

AI takes care of the simple calls—the quick questions, the routine issues. That’s the point.

But what’s left behind isn’t an easier job. It’s a more concentrated one.

Now most calls are the difficult ones. The angry customers. The complex situations. The ones that couldn’t be solved automatically.

The mix disappears. The natural rhythm of the work changes. There are fewer small wins, fewer resets, and a lot more sustained intensity.

From a system perspective, it’s working exactly as intended.

From a human perspective, the job feels very different.

This is the part that’s easy to miss if you’re only looking at outcomes.

We can measure efficiency. We can see improvements in output. We can track time saved and cost reduced.

What’s harder to see is what’s happening underneath all of that.

To people’s confidence in their own thinking.

To how work actually feels day to day.

To how much we trust not just the system, but each other.

At Fortrea, as we think about talent strategy, this isn’t theoretical.

We want to move faster. We should be using better tools. AI is going to be part of how work gets done.

But we also have to be intentional about what we don’t hand over.

Where people still need to sit with a problem before jumping to an answer. Where they have to work through ambiguity instead of being guided around it. Where they learn how to think, not just what to produce.

Because that’s where capability actually comes from.

The system works—until it doesn’t.

The model is right—until it isn’t.

And when that moment comes, what people actually understand—not what they can prompt for—starts to matter a lot.

We may be solving for efficiency at the system level in ways that quietly destabilize capability at the human level.

AI can absolutely accelerate development.

But only if we’re careful about what still needs to be built the long way.

Because not everything should be frictionless.

Some of the friction is where the learning is.