When in the early 1930s, Ralph Stackpole painted Industries of California inside San Francisco’s Coit Tower, work looked like this: rows of human bodies aligned around machines, hands synchronised, eyes focused, labour made visible through repetition and physical presence.

Those murals captured a truth of their time. Productivity was physical. Value came from muscle, coordination, and scale. Humans were essential—but largely interchangeable—components of industrial systems.

Nearly a century later, we are entering another transformation just as profound. But this time, machines are not replacing our muscles. They are absorbing something far more intimate: our cognitive work.

The future of work will not look like the murals. And that difference matters.

The truth we hesitate to say out loud

Let’s begin with what is often softened, avoided, or turned into emotional spectacle—especially on social media.

Yes—fewer people will be needed to perform many kinds of work.

AI systems are already reducing labor demand across entire categories:

• Coding
• First-draft writing and summarization
• Scheduling, coordination, and compliance
• Customer support triage
• Routine analysis and forecasting.

This is not a distant scenario. It is a structural shift already underway.

Every major technological transition follows a similar pattern. Industrial automation removed factory hands. Software flattened clerical hierarchies. AI now compresses layers of what might be called cognitive factory work—roles whose primary value lies in producing, transforming, or moving information from one format to another.

What I find myself wrestling with is not whether this reduction will happen—it will—but whether we are being honest about what replaces it.

Because the crucial distinction is this:

AI eliminates tasks. It does not eliminate work.

What disappears is not human contribution itself, but a particular shape of contribution—one that has quietly defined most modern jobs.

Technology doesn’t remove work — it redraws boundaries

Every technological shift redraws three boundaries at once:

1. What is automated
2. What is augmented
3. What becomes newly valuable.

AI is no exception. But it redraws these boundaries closer to the core of human cognition than any technology before it.

In the industrial era, work was organised around physical proximity. You had to be where the machine was. Your value was inseparable from your presence.

In the digital era, work reorganised around screens and networks. Presence became virtual, but sustained human attention remained central.

AI changes the organising principle again.

Machines no longer merely assist humans. They increasingly perform cognition autonomously—searching, drafting, summarising, classifying, recombining. This forces a boundary question we can no longer postpone:

What belongs to machines, and what must remain human?

The new dividing line: execution versus judgment

AI excels at:

• Pattern recognition at scale
• Speed, consistency, and availability
• Recombining existing knowledge
• Operating within defined objectives
• Optimisation under known constraints.

In short, AI is extraordinarily good at execution.

Humans, however, remain essential for a different class of work:

• Deciding which questions matter
• Defining goals, values, and trade-offs
• Interpreting meaning across contexts
• Navigating ambiguity and ethical tension
• Taking responsibility when outcomes have consequences.

AI can generate options.
Humans must decide what is worth choosing—and live with the results.

This is where the boundary truly sits. Not between artificial intelligence and human intelligence, but between execution and judgment.

And judgment, unlike computation, is neither evenly distributed nor easily scaled.

The hybrid zone: where the future of work actually lives

Most meaningful work in the coming decades will not sit cleanly on one side of this boundary.

It will live in what I think of as the hybrid zone.

In this zone:

• Humans set intent; AI explores possibilities
• Humans frame problems; AI accelerates search
• Humans interpret outcomes; AI generates variation
• Humans remain accountable; AI supplies leverage.

We already see this pattern emerging:

• Researchers working with AI to support medical diagnosis
• Policymakers using simulations to test scenarios before acting
• Public leaders relying on models to surface patterns and decide on priorities
• Creators shaping arguments while AI assists with drafts
• Managers steering systems rather than supervising individuals.

The nature of work shifts accordingly. Less time is spent producing outputs. More time is spent deciding what should be produced, why, and at what cost.

Value moves upward—from throughput to judgment.

A harder question: who gets to do the “human” work?

This is where the conversation becomes uncomfortable—but necessary.

As AI absorbs execution, access to judgment—not intelligence—becomes the new dividing line. And judgment is unevenly distributed, credentialed, and protected.

Who gets to operate in the hybrid zone?
Who is trusted to interpret, decide, and govern?
Who remains confined to execution until execution disappears?

The challenge ahead is not whether individuals can adapt. Many will.

The deeper challenge is whether our governments, businesses, and institutions are capable of redesigning roles, incentives, and authority around judgment rather than output.

Without that redesign, AI will not democratise work. It will stratify it.

Not dehumanisation, but re-humanisation

At first glance, this shift can feel like erosion. Less labour. Fewer roles. More automation.

But there is another way to see it.

The industrial murals show humans reduced to synchronized components—necessary, but replaceable. Their value lay in endurance and compliance.

AI, paradoxically, pushes human value in the opposite direction.

As machines absorb scale, speed, and repetition, what remains distinctly human becomes more—not less—important:

• Judgment under uncertainty
• Responsibility without delegation
• Creativity with consequences
• Moral and social accountability.

This helps explain a parallel cultural shift: the renewed interest in slow, bounded, deliberate forms of work—print journalism, physical books, intentional research.

These are not nostalgic retreats. They are counterweights.

AI is infinite, fast, and unbounded. Humans still need spaces that are finite, slow, and accountable.

The mural we are painting now

If the Coit Tower murals were painted today, they would not show rows of workers bent over machines.

They would show individuals in dialogue—humans interacting with systems, interpreting outputs, negotiating uncertainty, and making decisions whose consequences extend beyond the screen.

Fewer bodies in rows. More responsibility per person.

The question is no longer whether humans will remain in the loop.

The question is whether we are willing to accept the weight that comes with being there.