W. Edwards Deming

Most organizations approach technology as a solution to a problem.

A process is slow, so we automate it. Information is fragmented, so we centralize it. Decisions are inconsistent, so we standardize them. People are overloaded, so we introduce artificial intelligence.

Each of these actions may be reasonable. But they share an assumption: that the problem resides in a specific part of the organization and therefore a better tool, a better model or a better process can solve it.

W. Edwards Deming challenged this way of thinking. For Deming, the fundamental question is not which solution to apply, but what system is producing the result we observe.

A system is more than a collection of processes. It includes the structure, incentives, information flows, management practices, measurement methods and, above all, the way people are expected to behave. Change one element without understanding the whole, and the system often adapts to preserve its underlying logic.

This is why organizations frequently experience the same problems even after adopting new technologies. The symptoms change, but the system that produces them remains intact.

Artificial Intelligence will operate within this system. It will not stand outside it. It will learn from its data, reproduce its incentives and reflect its structure. It can amplify what works. But it can also amplify what does not.

Before asking what AI can improve, leaders must therefore ask a more difficult and more important question: What system are we asking it to improve?

Without a clear understanding of the system, we risk using extraordinary technologies to optimize the wrong things — or to accelerate outcomes that we do not actually want.

Measurement offers something highly attractive to leaders: clarity.

A number seems to tell us where we are, whether we are improving and what should be done next.

In complex organizations, this clarity can be seductive. It reduces uncertainty, simplifies discussions and creates the impression of control.

Deming did not argue against measurement. He argued against the misuse of measurement — particularly when numbers are treated as substitutes for understanding.

A metric can inform. It can guide. It can reveal trends. But a metric does not explain why something is happening. It does not describe the system. It does not capture the unintended consequences of a change.

When measurement is divorced from the system, it can create dangerous incentives. People begin to optimize the number instead of improving the outcome. Activity is confused with progress. Short-term results are rewarded while long-term capability is neglected.

Artificial Intelligence intensifies this risk. Its ability to generate data, dashboards and predictive indicators can make the organization feel more informed than it actually is.

But more information does not automatically lead to better judgment.

An AI model may identify patterns that were previously invisible. Yet if the underlying assumptions are flawed, if the incentive structure is distorted or if the system rewards the wrong behavior, the organization may simply scale its existing mistakes with greater precision.

Deming reminded us that measurement must serve a larger purpose: the improvement of the system as a whole, not the appearance of performance in its parts.

Used well, measurement can help organizations learn. Used poorly, it can produce a sophisticated form of illusion — the belief that what is measured is what matters, and that what is quantified is what is understood.

One of W. Edwards Deming's most important contributions was his insistence that most problems in organizations are not the result of individual failings, but of the system in which people work.

When outcomes are poor, the natural reaction is often to look for the person responsible. Someone made a mistake. Someone was not careful. Someone was not capable enough.

This instinct is understandable. But it is also dangerous. It directs attention away from the real question: what conditions produced this result?

People operate within a system of processes, incentives, information, resources and constraints. If the system is poorly designed, even capable and well-intentioned people will consistently produce undesirable outcomes.

Blaming individuals may create the illusion of accountability, but it does not improve the system. It often leads to more defensive behavior, less openness about mistakes and a reduced willingness to surface problems.

Deming argued that management has a fundamental responsibility to design and maintain a system that helps people succeed. This includes setting clear purposes, removing unnecessary obstacles, providing the right tools and creating an environment in which learning is possible.

In an AI era, this responsibility becomes even more important.

Artificial Intelligence can make systems more complex. It can introduce new dependencies, new types of risk and new forms of opacity. If the system is not well understood, people may be asked to work with tools and processes that they do not fully control or even comprehend.

When something goes wrong, it will be tempting to attribute the problem to the individual who used the tool. But the more responsible question is whether the system provided clear guidance, appropriate training, transparent information and realistic expectations.

Management must resist the simplicity of blame. It must undertake the more difficult task of understanding how the system works, where it is failing and what needs to change.

A culture of accountability is important. But it must be directed at the right level. The responsibility for the system lies with management. Without this responsibility, organizations risk automating their problems instead of solving them.

Artificial Intelligence offers an extraordinary opportunity to improve how organizations work, decide and serve their customers.

But technology is not a substitute for understanding. It can accelerate a system. It can extend its reach. It can increase its complexity.

It cannot compensate for a poorly designed system. It cannot replace the need to question assumptions. It cannot determine what truly matters for the organization, its people, its customers and society.

Before automating a process, leaders must take the time to understand it.

These questions are not a barrier to progress. They are what make progress sustainable. They help ensure that technology is applied to the right problems, in the right way and with the right expectations.

A clear understanding of the system may reveal that a process should be simplified before it is automated, that certain activities should be eliminated, or that the real opportunity lies elsewhere.

In some cases, the answer may be that automation is not the solution. Human judgment, experience and discretion may remain essential, especially in areas that involve uncertainty, trade-offs and human consequences.

In other cases, understanding the system may allow Artificial Intelligence to be applied far more effectively, with clearer objectives, better data, proper incentives and a stronger alignment with the organization's purpose.

Deming reminded us that improvement begins with knowledge. Without an understanding of the system, we risk automating the wrong things, measuring the wrong outcomes and reinforcing the very problems we hoped to solve.

Technology will continue to evolve. Its capabilities will expand. But the fundamental discipline remains the same.