Part I: The bug in the machine

This is part I of “Rethinking work for the intelligence era,” a two-part series exploring why organizations struggle to turn insight into action—and how they can close that gap.

Most organizations can see change coming, but few can move fast enough to do anything about it. That gap—between knowing and doing—is the defining operational challenge of the AI era, as organizations see more than they can act on.

But closing this gap isn’t primarily a technology problem. It’s a design problem: how work is structured, where decisions are made, and whether intelligence can move fast enough to shape what happens next.

Here’s the paradox: we spent over a century designing industrial systems to make humans behave like machines. We perfected it. The systems we inherited were optimized for efficiency—speed, consistency, scale. They worked brilliantly in stable environments. But that success came with a hidden cost.

You are what you practice. In optimizing for efficiency, we weakened our ability to interpret, decide, and adjust. Now we’re forced to confront the trade-off. As efficiency becomes automated, the advantage shifts to what we set aside.

The question is no longer just how to produce more—but whether organizations can turn what they know into action fast enough to keep up with change. That requires intelligence to move—where signals are sensed, decisions are made, and action happens.

We’ve designed work as if humans were machines: constant output, infinite availability, instant response. We call emotional suppression “professionalism” and reward endurance over recovery.

We optimize calendars, compress timelines, and measure time as if it were perishable inventory, expiring on a shelf. We’ve normalized this mindset so thoroughly that it barely registers. Sleep signals inefficiency. Doubt seems weak. Fluctuation looks unreliable.

Is humanity the bug in the machine?

For most of industrial history, the answer seemed to be yes.

In the 1890s, Frederick Taylor studied how workers moved in factories. He timed each motion with a stopwatch, identified the fastest workers, broke their movements into steps, and trained everyone to do it the same way. The goal was to turn variable human behavior into predictable machine-like performance. Productivity doubled, then tripled.

Henry Ford took it further. His assembly line cut the time to build a car from 12 hours to 93 minutes. The pattern spread everywhere: study the best way, write it down, remove worker choice, scale it. And it worked. 

The appeal of industrialization was clear. Human judgment was slow and unreliable. It introduced risk. Machines were different. They didn’t get tired. They didn’t need motivation. They did exactly what they were designed to do, every time.

So we designed organizations to work like machines. We separated thinking from doing, making from learning, strategy from execution—assuming each would perform better in isolation. We built systems that controlled everything, breaking apart the natural way humans sense, make, and learn.

When you treat people like machine parts long enough, you start to believe they are. Replacing humans with actual machines stops feeling radical. It feels logical. In a system designed like a machine, anything unpredictable looks like a defect.

But here’s what we missed: efficiency was never the source of competitive advantage. It was the amplifier.

Breakthroughs didn’t come from doing things faster. They came from experimentation and learning—the human abilities that industrial systems were designed to remove. Taylor’s methods increased productivity, but someone still had to invent what was being produced.

Efficiency multiplies success once it exists. It can’t create it.

Industrial systems optimized for replication—doing the same thing faster and cheaper. That worked when the world was stable. But when conditions change, replication becomes a liability. The systems that once made organizations powerful are now making them fragile.

How intelligence was fragmented

Before industrialization, a craftsperson sensed, made, and learned in one motion. They observed the material, shaped it, and adapted accordingly. Knowledge and creation were inseparable.

Sense–Make–Learn. Repeat.

Sensing means noticing and interpreting signals: changes in context, environment, and perception. Making means responding through action: shaping, exploring, and executing. Learning emerges from observing results and updating understanding.

This cycle strengthens intelligence: action sharpens perception, and feedback improves judgment—linking what we see with what we do.

Industrialization broke that loop to scale. Sensing moved to analysts, making moved to workers, and learning moved to managers. The system became more efficient, but more fragmented.

What we gained in efficiency, we lost in intelligence.

Toyota is the exception that proves the point. Any worker on the production line can stop the entire operation the moment they detect a problem. The sense–make–learn cycle remains unbroken across roles, making Toyota one of the most consistently high-quality and adaptive manufacturers in the world.

Most organizations don’t grant workers this authority—not from malice, but because the system wasn’t designed for it. Fragmented intelligence requires centralized control to maintain consistency. Giving frontline workers decision-making power feels like it threatens standardization and predictability. So the capability remains locked away.

When efficiency breaks

For a long time, that trade-off worked because the environment allowed it. Markets moved slowly enough for plans to be made, and organizations had time to optimize for stability. Direction could be set centrally and executed predictably.

Today, the conditions are different. Technology cycles are shorter, information moves instantly, and competition can emerge from anywhere. Environments now shift faster than industrial organizations were designed to respond.

Kodak is the clearest example of what happens when that breaks down. Their own engineers invented digital photography in 1975. But the organization had optimized for film, not for acting on new ideas. By the time the structure allowed a real response, the market had already moved. What Kodak experienced slowly, organizations now face at speed.

AI is perfecting exactly what industrial systems were built to optimize: efficiency at scale. It scales accumulated knowledge, routine reasoning, and established processes at extraordinary speed and with precision.

If the game is efficiency, machines will outperform humans. The real question is whether efficiency is still the game that matters most.

Industrialization rewarded efficiency, so we optimized for predictability over adaptability, specialization over integration, and throughput over understanding. Now, machines do that better than we ever could.

The intelligence era isn’t fundamentally about AI. It’s what AI reveals: that the qualities industrial systems suppressed—judgment, reflection, contextual sensitivity—are now the source of advantage.

Production is no longer the primary constraint. The harder challenge is deciding what to make, why it matters, and how to adjust as conditions change. The central task of organizations has shifted from executing known solutions to discovering what to do next.

The real advantage now is adaptability: the practical expression of intelligence in changing environments. It’s the ability to interpret ambiguity, exercise judgment under uncertainty, respond in real time, and form new connections and ideas.

Humanity was never the bug in the machine. The flaw was a system designed for efficiency at the expense of the intelligence required to adapt.

Living systems, not machines

To understand why adaptability matters, it helps to look at how intelligence actually develops in living systems.

In living systems, intelligence isn’t stored and retrieved; it develops through continuous interaction with the environment. No central authority decides how a cell responds. Every part perceives, reacts, and adapts. That isn’t a metaphor for good organizational design. It’s the underlying principle that organizational design either works with or against.

Living systems adapt through variation and feedback. Performance improves not through constant exertion, but through learning. Variation isn’t a flaw—it’s how adaptation happens. Energy rises and falls. Stress is followed by repair. Growth alternates with consolidation. These are not mechanical traits. They are features of intelligence.

Industrial organizations were designed as machines, with intelligence concentrated in a few roles. In living systems, every cell perceives, responds, and adapts. In environments defined by constant change, intelligence cannot sit at the top. It must be distributed throughout the system.

Organizations learn through the same underlying dynamic. The challenge isn’t recognizing signals, but turning what they recognize into coordinated action. When perception flows directly into action, which flows into learning, capability strengthens. When they’re separated, capability weakens: the organization sees clearly but moves slowly.

That’s the design problem.

Redesigning work for intelligence

Industrial systems optimized efficiency by fragmenting intelligence. Signals travel up through management layers. Decisions travel back down. The feedback loop breaks in between.

When teams detect signals—shifting customer behavior, emerging technologies, new competitors—action doesn’t follow directly. By the time a response returns, the problem has changed. The delay doesn’t just slow response—it disconnects sensing from making from learning.

Over time, people stop trusting the system. Why sense what won’t be acted on? Why propose what won’t be heard? The gap between knowing and doing erodes both adaptability and trust.

Adaptive systems create advantage by integrating intelligence—pushing decisions to where sensing occurs.

Amazon built this deliberately. Small teams—“two-pizza teams”—own their decisions end to end. There’s no approval chain between detecting a signal and responding to it. When the people closest to the work can sense and decide without waiting for permission, the feedback loop closes immediately.

This isn’t about empowerment as culture or adopting agile practices. Both operate within existing structures. The challenge is the structure itself—whether an organization is designed to let intelligence move at all.

The defining operational challenge of the AI era is the distance between knowing and doing—whether organizations can act on what they know before the environment changes. Closing it requires redesigning work around how human intelligence actually functions.

The systems we inherited asked humans to behave like machines. Now, machines have taken over that role. The advantage shifts to organizations that design for human intelligence—where judgment, creativity, and learning are embedded in the flow of work rather than layered atop. 

Part II, coming in April, examines how organizations can close this gap—redesigning structures so that insight moves faster, action follows directly, and learning continuously reshapes what happens next.