Stop Asking What AI Will Replace. Start Asking What People Can Build.

“Which jobs will AI replace?”

It may be the most frequently asked question about the future of work.

We hear it at conferences, in boardrooms, during workforce planning discussions and in conversations between colleagues.

It is a legitimate question.

AI will change jobs. Some tasks will disappear, others will be redesigned, and new roles will emerge. Organizations have a responsibility to examine the implications honestly.

But I have been thinking about another question.

What if, instead of telling employees how AI will change their work, we invited them to help design that change?

Recently, at an event focused on AI and the future of work, I heard an example that stayed with me.

Not because the technology was particularly spectacular.

But because one company had chosen a different starting point.

It started with its people.

1. They started with a question. Not an automation plan.

In many organizations, AI transformation follows a familiar sequence.

Leadership identifies opportunities for automation.

Technology teams evaluate solutions.

Finance calculates potential savings.

Projects are approved.

Employees are informed about what will change and trained to use the new tools.

There is nothing inherently wrong with this approach. Technology implementation requires structure, investment decisions and clear accountability.

But the company described at the event did something different.

It invited employees to identify the opportunities themselves.

The challenge was simple:

“Look at what you do every day. What could we automate? What could we simplify? What could we do faster — or better — with AI?”

Ideas started coming from teams.

People met, discussed possibilities and helped establish priorities.

Groups of builders and users were formed.

Some employees worked on developing solutions.

Others tested them, provided feedback and assessed whether they actually improved everyday work.

I don't have independently verified performance results from this company's initiative, so I wouldn't claim that participation alone made the program successful.

But one thing struck me.

Employees were no longer simply the people expected to adapt to change. They were helping build it.

And that is a fundamentally different leadership approach.

2. The people doing the work often know where the real problems are

A CEO sees strategy.

A CFO sees costs.

A CIO sees technology and infrastructure.

But the person doing the work every day sees something else.

The report nobody reads.

The information entered three times into different systems.

The approval that delays a decision for two weeks.

The repetitive task that takes 40 minutes but could potentially take five.

The process everyone follows, even though nobody remembers why it exists.

And perhaps most importantly, the difference between how work is supposed to happen and how it actually happens.

This knowledge is valuable.

Yet organizations don't always capture it systematically.

A top-down AI program may identify technically attractive opportunities while overlooking the everyday friction that consumes employees' time.

When people help identify problems, different opportunities emerge.

Not necessarily the most sophisticated.

Sometimes the most useful.

If we want AI to transform work, why wouldn't we start by listening to the people who understand that work best?

That doesn't mean employees should independently decide which technologies to purchase or which processes to automate.

We still need technical validation, data security, financial analysis, risk management and clear decision-making authority.

But if we define problems exclusively from the top, we risk building excellent solutions to the wrong problems.

3. There is a difference between using AI and having a say in how it is used

Imagine two organizations.

In the first, employees receive an announcement:

“Starting next month, we are introducing a new AI tool. Training will be mandatory.”

In the second, the conversation begins differently:

“We want to rethink how we work. Tell us which activities consume unnecessary time, which problems you encounter and what you believe we could improve.”

Both organizations may eventually implement the same technology.

But the employee experience is different.

In the first, people receive a change.

In the second, they are invited to contribute to it.

I see this as the difference between AI adoption and ownership of change.

Adoption asks:

Will people use the technology?

Ownership asks:

Do people understand the problem? Can they influence the solution? Do they know what outcome the organization is trying to achieve?

Ownership doesn't mean everyone becomes a developer.

It doesn't mean every idea gets implemented.

It means participation is real rather than symbolic.

Research on employee voice, including Elizabeth Morrison's review of the field, helps explain why employees choose to share ideas or remain silent.

Amy Edmondson's research on psychological safety also highlights the importance of environments where people can ask questions, acknowledge uncertainty and learn together.

These studies don't prove that employee involvement guarantees successful AI implementation.

But they support a broader principle.

People are better positioned to contribute when they have a genuine opportunity to question, learn and influence.

4. When people don't know what's coming, they start imagining scenarios

Consider how an employee might interpret this announcement:

“Our company is launching a major AI automation program.”

What does it mean?

Will their work become easier?

Will they need new skills?

Will their role change?

Or could their position disappear?

When information is missing, people fill the gaps themselves.

And when their livelihood is involved, those assumptions are not always optimistic.

I don't believe leaders should promise that AI will never affect employment.

In many organizations, that would be an irresponsible promise.

But leaders can explain what they are trying to achieve, what is already known, what remains uncertain and how decisions will be made.

They can also create meaningful opportunities for employees to contribute.

Because there is a difference between saying:

“AI is coming. You will need to adapt.”

And saying:

“AI is changing how we work. Let's identify together what is worth changing.”

The second approach doesn't eliminate uncertainty.

But it can turn some of that uncertainty into participation.

And that matters.

5. The most important question comes after automation

Now imagine that AI implementation works.

A task that previously required ten hours can now be completed in four.

Six hours have been saved.

Excellent.

What happens to those six hours?

Do we use them exclusively to reduce costs?

Do we fill them with additional reporting?

Do we simply expect employees to produce twice as much?

Or do we redesign the role so people can spend more time on activities that create greater value?

Deeper analysis.

Better customer conversations.

Complex problem-solving.

Creativity.

Judgment.

Better decisions.

This is where I believe one of AI's biggest opportunities lies.

Not simply doing the same things faster. But enabling people to do more valuable things.

Because time saved is not automatically value created.

6. The productivity trap: faster doesn't necessarily mean better

Imagine two companies.

Both introduce AI into their reporting processes.

Both reduce the time employees spend preparing reports.

In the first company, employees are expected to produce more reports.

In the second, leaders ask a few additional questions.

Do we still need all these reports?

Who uses them?

Which decisions do they support?

What information is missing?

How could the time saved be redirected toward better business analysis?

The first company has accelerated activity.

The second has started to rethink the value of that activity.

This is a hypothetical example, but the distinction is important.

I think of it as the Capacity-to-Value Gap.

The distance between the capacity AI releases and the measurable business value an organization creates from it.

This is a practical framework I propose, not an established scientific metric.

AI can release capacity.

But leadership determines what that capacity becomes.

And sometimes the most important decision isn't what to automate.

It's what to stop doing altogether.

7. Builders and users are not enough. We need a third group.

I liked the idea of organizing employees into builders and users.

But I would add a third group.

Decision owners.

People who are accountable for the business decisions.

Because a good idea isn't enough.

A functioning technological solution isn't enough.

Even a successful pilot isn't enough.

Someone must decide whether the solution should be scaled.

Whether the benefits justify the investment.

Whether the risks are acceptable.

Whether the process itself needs redesigning.

Whether roles and responsibilities need to change.

And how success will be measured.

I would structure the approach around five responsibilities:

Employees identify the problems.

Builders develop potential solutions.

Users test and improve them.

Decision owners determine what should be implemented and scaled.

Leadership redesigns work around the new possibilities.

Without that final step, organizations risk creating dozens of interesting AI pilots while changing very little about how the business actually operates.

8. If I were leading an AI transformation program tomorrow

I wouldn't begin by asking every department for a list of activities that could be automated.

I would begin with five questions.

1. Which activities consume time without creating sufficient value?

Not every process needs automation. Some need simplification. Others should disappear.

2. Which problems do employees know about that never reach the leadership agenda?

Important opportunities may be hidden there.

3. What can employees genuinely help build or test?

Participation requires time, capabilities, support and clear responsibilities.

4. What will we do with the capacity we release?

If we cannot answer this, we may have an efficiency project — but not necessarily a transformation of work.

5. What should these roles look like in two years if our AI initiatives succeed?

This is the question that connects AI to Human Capital, organizational design and business strategy.

Because we cannot transform activities without reconsidering roles, capabilities, accountability and decision-making.

9. Perhaps the biggest transformation isn't technological

The World Economic Forum's Future of Jobs Report 2025 projected that approximately 39% of workers' existing skill sets would be transformed or become outdated between 2025 and 2030.

Microsoft's 2025 Work Trend Index also described growing expectations among leaders regarding the integration of AI agents into business operations.

These are forecasts and reported expectations, not guaranteed outcomes for every organization.

But the direction is difficult to ignore.

Technology is forcing us to reconsider how work is organized.

And this is where leadership choices become visible.

We can use AI to accelerate existing processes.

Or we can use it to question whether those processes still make sense.

We can introduce new tools into organizations built around old assumptions.

Or we can rethink work, responsibilities and value creation.

We can treat employees primarily as a cost to optimize.

Or recognize them as a source of knowledge, ideas and transformation capability.

These choices are not always mutually exclusive. Organizations must manage costs, performance and people.

But how leaders balance those priorities says a great deal about the kind of organization they are building.

10. The question I took away from that event

Of all the perspectives on AI I heard that day, what stayed with me wasn't a technological demonstration.

It was a question asked of employees:

“What could we do better with AI?”

I think that's a remarkably powerful starting point.

And I would take it one step further.

If AI gives our people more time, what will we enable them to build with it?

More reports?

Or better decisions?

More activity?

Or more valuable outcomes?

More pressure to increase productivity?

Or more room for thinking, initiative and problem-solving?

I don't believe the answer will be the same for every company.

But I believe every leadership team should be having this conversation.

Because AI will undoubtedly change how we work.

Whether people become merely the recipients of that change — or help build it — remains a leadership choice.

And perhaps instead of obsessively asking which jobs AI will replace, we should start asking:

What could our people achieve if we removed unnecessary work and gave them more room to create value?

That's the AI conversation I'd like to hear more often.