The Productivity Illusion

During my years as a Flight Engineer, I learned something that has stayed with me long after leaving the cockpit: more information doesn't necessarily mean better decisions.

Inside a helicopter, you're surrounded by instruments, indications and procedures. Their purpose isn't to produce more information, but to help the crew understand what matters and make the right decision when it counts. Completing a checklist is important, but it's not the achievement. Establishing that the aircraft is safe and ready is.

Years later, watching organizations embrace artificial intelligence, I began noticing a familiar problem. We are becoming remarkably good at producing information, reports, analyses and recommendations. Yet we rarely stop to ask whether all that activity is improving the decisions we're supposed to make.

Then I came across a study that captured this disconnect almost perfectly.

In 2025, researchers at METR studied sixteen experienced software developers working on real open-source projects. Before the experiment, the developers expected AI tools to make them 24% faster. Afterward, they still believed AI had made them 20% faster. The measured result was surprising: their tasks had taken 19% longer.

What caught my attention wasn't simply the slowdown. It was the distance between what the developers believed they had achieved and what the measurements showed. The experiment examined a specific group using early-2025 AI tools, so its findings shouldn't be generalized to every workplace. But it raised a question that extends far beyond software engineering:

Are we becoming more productive, or simply getting better at producing the evidence that makes us feel productive?

The perception gap between Speed & Execution.

Expected: 24% faster. Perceived: 20% faster. Measured: 19% longer. Source: METR, July 2025.

When output becomes the illusion

A presentation takes minutes. A report appears before we've fully understood the question. An email is written, rewritten and polished in seconds. The screen fills, the task list shrinks and we feel productive. Yet producing something and improving something are fundamentally different events.

Consider a marketing team that generates twenty campaign concepts without discovering which one customers actually want. Or a manager who receives ten polished summaries but continues postponing the decision they were supposed to inform. These are hypothetical examples, but they illustrate a genuine problem: when output becomes easier to create, we start treating output itself as the achievement.

The danger is subtle. We count reports, presentations, messages and completed tasks because they are visible and easy to measure. Their actual contribution to better decisions, customer experiences or business results is much harder to establish. And when production becomes almost effortless, the temptation to confuse activity with progress grows stronger.