COLUMN – In this column, different lean and technology experts answer the same pressing question shaping today’s tech/AI debate. This month, the question on everyone's mind: will AI really take our jobs?

Words: Eivind Reke and Sandrine Olivencia

THE QUESTION

According to a recent Financial Times article, AI may not have led to widespread job losses yet, but it is quietly reshaping work—especially slowing hiring in AI-exposed and entry-level roles. Rather than eliminating jobs, it is redistributing tasks while creating new, less visible work in oversight and coordination. The result is apparent efficiency gains, but uncertain system-level productivity.

Across society, predictions about the impact of AI range from jobs apocalypse to just another tech fad, which begs the question: can Artificial Intelligence really replace workers?

THE ANSWERS

We have seen this pattern before, with the hype around Industry 4.0 for example. New technologies promise transformation, but rarely eliminate work—instead, they reshape it. So, no… AI won’t replace workers, but it can transform the way value is created when it’s combined with Lean.

The adoption of AI seems to follow different patterns in different organizations. In some, it mirrors earlier Industry 4.0 initiatives: top-down, technology-driven, and focused on efficiency gains. Advanced analytics, automation, and algorithmic decision-making promise faster processes and reduced costs. In other organizations, it seems more bottom-up, with software engineers experimenting with LLMs to automate code writing. The latter approach is arguably more aligned with Lean, which teaches us that performance does not come from tools alone, but from people development through experimentation, problem solving, and kaizen.

Regardless, implementing AI technologies, such as LLMs, in manufacturing is hard work, because the model is not trained on the data it will be asked to analyze. When AI is introduced without considering how it supports (or disrupts) this dynamic, its impact is most likely limited. The Financial Times observes that this is reflected in the fact that hiring is slower in AI-exposed roles: certain tasks disappear or are automated, but the organization still requires people to interpret outputs, manage exceptions, and ensure that systems function as intended. In other words, you need people who deeply understand the technical conditions of the specific manufacturing challenge on hand.

In manufacturing, digital tech can improve visibility, provide real-time data and insights that make problems easier to detect. AI can accelerate analysis, identify patterns, and even suggest improvements. It cannot replace the core of kaizen—the human capability to question, experiment, and learn—but it can change its process. Operators and engineers spend less time gathering and processing data, and more time understanding it, coordinating across functions, and implementing improvements. The upside is that more data can be analyze faster; the downside is the sensation called “feeling of knowledge”. Even if we strongly believe we know something, it might not reflect reality.  At the same time, new types of work emerge: maintaining data quality, validating algorithmic outputs, and aligning digital systems with operational realities. Research consistently shows that digital technologies alone do not guarantee performance improvements. In fact, without alignment with lean principles and strong employee involvement, their impact can be uneven or even negative. The real gains emerge when technology and human systems are integrated. When AI enhances the ability of people to identify waste, solve problems, and improve safety, quality and flexibility to create better flow.

The idea that AI will replace workers is based on a misunderstanding of how work systems function. Jobs are not simply collections of tasks that can be automated independently; they are part of broader system where value is created through the interaction between people, processes, and technology. AI can change these interactions, but it can’t eliminate the need for human contribution. If anything, it makes it more critical as AI, unlike people, doesn’t possess any knowledge on its own. From a Lean perspective, the path forward is clear. AI should not be seen as a substitute for people, but as a tool to enhance people’s ability to improve work. The real opportunity lies in combining the strengths of both: the speed and scale of AI with the creativity, judgment, and learning capability of humans.

Artificial Intelligence is unlikely to fully replace workers in the way many predict. What we are seeing instead is a recomposition of work: AI can automate specific, well-defined tasks, especially those that are routine, repetitive, and easy to codify. This shifts the role of humans from simply executing tasks to understanding problems and making decisions.

I was recently with ITU’s Head of Service Desk. We were trying to build a simple performance board for quality, lead-time, stock, and productivity using data from their ticketing system and AI. We spent time refining prompts, but we struggled to get Claude and Copilot to generate the “right” indicators, not really because we didn’t know how to write prompts but because we didn’t know how the system actually works.

AI cannot compensate for a lack of understanding of how things actually work, but it certainly exposes it. Across organizations, this technology is accelerating execution, but it is also revealing something deeper: most organizations are not designed for people to understand the systems they operate; they are designed to act, to deliver, and to move work forward. For years, this was sufficient because execution was the main constraint, but with AI execution becomes cheap and what becomes scarce is understanding.

This is where the confusion begins: faced with inconsistent outputs or questionable insights, people often focus on improving prompts, but better prompts only matter if there is clarity. Without a solid grasp of how the system and the world around it work, better prompts simply produce more convincing answers built on the same flawed assumptions. Vague thinking leads to useless AI output, which leads to bad decisions made faster.

When Toyota developed the Toyota Production System, the goal was never primarily efficiency or cost reduction. Its core idea is to “build” people who understand the system they operate, which means going to the gemba, making problems visible, understanding cause and effect, learning through experimentation, and taking responsibility for improvement. Lean is not about optimizing isolated tasks but about developing the capability to see, think, and act on the system as a whole.

This is what AI now demands, because it has no real understanding of the real world; it generates outputs based on patterns, not meaning, so it cannot decide what matters, ensure coherence, or arbitrate real-world trade-offs. Those responsibilities remain entirely human.

Yet, most organizations are structured in ways that make this difficult: silos fragment knowledge, handoffs obscure accountability, and local optimization hides system-level effects, so people are trained to execute tasks and comply with processes rather than to understand how those tasks connect into a coherent whole.

Lean forces a different question: how does the system create value, and where does it fail? When people begin to see end-to-end flows, understand trade-offs, and connect decisions to outcomes, something changes: our understanding widens. AI is pushing organizations into a new reality: from executing tasks to framing problems, from following processes to understanding systems, from optimizing locally to optimizing globally, and from relying on experience to learning continuously. Lean is the only widely proven system that structures learning, embeds it in daily work, and scales it across teams. Without Lean, AI leads to superficial productivity gains accompanied by more rework, more misalignment, and greater fragility; with it, AI becomes a powerful lever for deeper learning.

‍