When technology changes, does the work?

AI is already reshaping how people approach their work. 

That much is clear. Employees use AI to draft documents, analyze information, summarize meetings, research questions, create content, and automate day-to-day tasks. Organizations are spending on AI tools and training people to use them. 

But let’s pause and ask a bigger question, one that sometimes gets overlooked: what actually happens to the work itself once AI becomes part of it? 

Because simply adding AI to a job isn’t the same as thoughtfully redesigning it. 

The risk is that organizations move quickly to implement new technology without stopping to consider what else may need to change.

New Technology, Old Work 

A common approach to using AI is straightforward: find a task that takes too long, bring in a tool to speed it up, train employees, and move on.

It makes sense. But work is rarely just a set of separate tasks.

After all, every part of an organization is connected. Decisions move from one role to another. Managers review their teams’ work, and employees often rely on information from other areas of the organization.

Change one part of that system, and you set off a ripple effect. 

Imagine, for instance, a team that used to spend hours preparing a report. Now, that same team uses an AI tool to deliver a first draft in mere minutes. 

At first glance, that sounds like an efficiency win. But what really happens next? 

Who reviews the AI’s output? What does the reviewer need to know in order to recognize an error? Who decides whether the information is reliable enough to use? Does the person who used to prepare the report still need the same skills, or new ones? Does the manager still need the same information? And if the report is wrong, who’s accountable? 

The technology may have changed how a task is carried out, but the implications run deeper because the workflow has changed too. Have we noticed?

The Work Is Changing 

We tend to frame the AI discussion around jobs: which ones will disappear? Which will be replaced? Which are safe?

Those questions are understandable, but they can distract us from changes happening much closer to home, particularly in the way work gets done.

For example, a project team that uses AI to generate first drafts, analyze information, or identify potential issues may need to rethink who reviews the work, who makes the final call, and how responsibility moves from one person to the next.

Or, an accountant who spends less time processing information can spend more time interpreting it and helping the business understand what the numbers mean.

None of these people necessarily have a completely different job title, but their work and what’s expected of them may look quite different. Their responsibilities can shift, the skills they need may evolve, and the way their performance is measured may need to be reconsidered. Even how they work with colleagues can be affected.

This is why the question shouldn’t just be: “How can we use AI in this role?” 

We also need to ask: “Now that AI can do some of this work, what should this role actually be responsible for?” 

Adoption vs. Redesign 

There is nothing wrong with introducing AI into an existing process. Sometimes that is exactly what an organization needs. However, the problem starts when we assume that adding a new technology is enough.

Recent research supports this. Deloitte’s 2026 Global Human Capital Trends research found that 59% of organizations are taking a technology-focused approach to AI, while only 6% of leaders say they are leading in the intentional design of human-AI interactions.

That should make us pause.

When AI becomes part of a process, it can affect more than just the task at hand. Responsibilities can shift, and decisions may end up being made at different points than before. Sometimes, the way work moves from one person to another needs to be rethought.  

Employees may find that some of the skills they relied on before are no longer enough, while others become more important. They may also need greater clarity about when their own judgment matters most. 

If we don’t stop to look at those things, we can end up with a new technology sitting inside an old way of working. 

That’s where the real opportunity may be missed. 

Adopting AI is about bringing technology into the organization. Redesigning work is about deciding what the organization should look like with that technology in it. 

There is another side to this conversation that often gets missed. Not every process needs to be reinvented just because AI is now available. In many cases, the most practical move is to use these new tools to make existing work faster or easier to manage. The real challenge is recognizing when technology has changed the work enough that the old approach no longer makes sense. 

To answer that, leaders need to look past the technology itself and pay attention to what’s happening day to day. How is the work actually getting done? Who is responsible for each step? Are the right people making decisions with the right information? Do handoffs between teams still work smoothly, or have they become points of confusion? Are employees being asked to learn new skills, or stop doing tasks that used to matter? And what do we do when something goes wrong? 

It’s easy to overlook these details when everyone is focused on buying the next AI platform. But these questions determine whether the organization is actually adapting or simply layering technology onto old habits. 

Work redesign is often treated as a technology exercise. But once AI starts affecting roles, responsibilities, decision-making, performance expectations, or collaboration, it becomes an organizational issue. That can mean revisiting job descriptions and performance measures, developing new skills, changing how teams collaborate and where decisions are made, or even questioning whether certain responsibilities still belong in the same role or department. 

And this is where AI connects directly to organizational development.  

Organizations have spent years talking about becoming more agile and adaptable. AI is putting that ambition to the test.  

It’s tough for any organization to truly adapt if technology moves forward but the way people work stays the same. When roles, routines, and expectations never shift, you may end up with new tools while all the old problems remain. 

Start with the work  

It’s easy to get caught up in the excitement of new technology, but sometimes the best question to ask is much simpler: What should our work look like right now? This isn’t about waiting for the perfect moment or for everyone to become AI experts. It’s about looking at what’s needed today. 

Taking a hard look at the work means asking where people are still essential, where AI can help, and when human judgment matters most. It involves noticing whether handoffs between teams still work, whether certain skills are becoming more important, or whether some tasks no longer fit the same role. Above all, it means thinking about what these shifts mean for the people who do the work every day. 

You won’t get a tidy AI implementation plan from these questions. Asking them, however, can help leaders see more clearly how the organization may need to operate as the work changes.  

When a company adopts AI, it’s usually pretty obvious. There might be a new tool, some training sessions, and people start using it. The harder thing to spot is whether the organization is actually changing how work gets done. 

Redesigning work is a more subtle process. It asks leaders to look closely at old habits and assumptions about how things have always been done. This is often a tougher conversation, but it is also more important. 

AI is already changing the work. The organizations that will truly benefit are not just the ones with the latest tools. They will be the ones willing to look closely at how work is organized and ask whether it still makes sense.

When work changes, organizations have a choice: fold the new technology into the old routine or take a closer look at whether the way the work is organized still makes sense.