• The workers who adopted AI most enthusiastically are burning out fastest. The technology that was supposed to give you your time back is taking more of it than ever.
• The paradox
• The numbers
• The adoption trap
• The structural ratchet
• What the Intensity Paradox means

§The workers who adopted AI most enthusiastically are burning out fastest. The technology that was supposed to give you your time back is taking more of it than ever.

For eight months, researchers from UC Berkeley and Yale embedded themselves inside a 200-person tech company that had fully adopted AI tools across engineering, product, design, research, and operations. The study, published in Harvard Business Review in February 2026, was designed to answer a simple question: what happens when a real company actually uses AI the way the brochures promise?

The answer was the opposite of the brochure.

Workers who adopted AI tools didn’t work less. They worked more. They worked faster, took on broader scope, and extended their hours — voluntarily. Product managers started writing code. Researchers took on engineering tasks. Designers began doing data analysis. AI didn’t eliminate tasks. It eliminated the friction that kept people inside their roles.

The researchers called it “workload creep.” Over eight months, employees absorbed responsibility for work that used to belong to other teams. The AI made starting a new type of project so easy that saying no felt irrational. If the tool can help you write the code, why wouldn’t you write the code? If the prompt can draft the analysis, why wouldn’t you draft the analysis?

The result was a workforce doing more kinds of work, at higher speed, across longer hours — without anyone asking them to. Burnout hit 62% of associates and 61% of entry-level workers, versus 38% among the C-suite executives who mandated the adoption.

The people who burned out fastest were the most enthusiastic adopters.

§The paradox

This finding has a name. Call it the Intensity Paradox — the tendency for productivity tools to increase the intensity of work rather than reduce it, with the highest cost falling on the people who adopt them most willingly.

The Intensity Paradox is not new to AI. It is a recurring pattern in the history of labor-saving technology. The washing machine didn’t give housewives free time — it raised the standard of cleanliness until people were washing clothes twice as often. Email didn’t reduce communication overhead — it multiplied the number of messages anyone was expected to process in a day. Smartphones didn’t create leisure — they erased the boundary between work hours and personal hours.

Every tool that reduces the effort per task increases the number of tasks expected. The savings are captured not by the worker but by the system.

AI is following the same pattern, with one critical difference: the speed. Previous productivity tools took years or decades to ratchet up expectations. AI collapses the cycle into months. The Berkeley study documented the full arc — adoption, scope expansion, role blurring, burnout — in under a year.

§The numbers

The Intensity Paradox is appearing across every dataset that looks for it.

The PwC 2026 Global CEO Survey of 4,454 executives across 95 countries found that 56% of CEOs report getting “nothing out of” their AI investments — no revenue gains and no cost benefits. Only 12% report that AI delivered both. The technology is making individuals do more work without making organizations measurably more productive. The effort goes up. The output stays flat.

Meanwhile, 85% of workers surveyed say AI saves them one to seven hours per week. But 37% of those savings are lost to rework — correcting AI errors, rewriting AI output, verifying AI claims. The net time savings are a fraction of what the tools promise, and the time they do free gets filled immediately with new work.

The result is a workforce that feels more productive while becoming more exhausted. The subjective experience of AI-assisted work is speed and capability. The objective experience is cognitive overload and ambient work — what researchers describe as sending “one last prompt” before lunch, running AI tasks during evenings and weekends, losing the recovery function of downtime.

Downtime used to be downtime. Now it’s just time you haven’t prompted yet.

§The adoption trap

The Intensity Paradox contains a particularly cruel mechanism: it punishes enthusiasm.

The workers who resist AI tools — who refuse to learn the prompts, who stick to their defined roles, who maintain their boundaries — are protected from the intensity ratchet. They do the same work at the same pace. Their managers may be unhappy with their adoption metrics, but their cognitive load hasn’t changed.

The workers who embrace AI — who learn every tool, who volunteer for cross-functional work, who demonstrate that they can do three jobs with the help of a prompt — become trapped. They can’t stop doing three jobs once they’ve proven they can. The AI didn’t give them back three jobs’ worth of time. It gave their employer three jobs’ worth of output from one salary.

This is why the burnout gradient runs in the wrong direction. The most enthusiastic adopters carry the highest load. The most skeptical workers carry the lowest. The system rewards adoption with more work and punishes it with burnout, while rewarding resistance with stability.

The cruel twist: companies like Block, Meta, Amazon, and Google are now tracking individual AI usage in performance reviews. In February 2026, Block simultaneously mandated daily AI use, tied AI fluency to performance evaluations, and laid off 10% of its workforce. The message is explicit: adopt the tool that’s being used to justify eliminating your colleagues, or get fired for insufficient innovation. Workers face a double bind where both adoption and non-adoption carry existential risk — but only adoption carries the burnout penalty.

§The structural ratchet

The Intensity Paradox operates through a ratchet that only turns one direction.

When AI is introduced, the initial experience is genuine relief. Tasks that took hours take minutes. Work that required coordination can be done solo. The first week of AI adoption feels like a superpower.

By the second month, the freed capacity is filled. New projects appear. Scope expands. Timelines compress. The baseline of expected output shifts upward.

By the sixth month — the point at which the Berkeley study found burnout emerging — the original workload has been replaced by a heavier one. But the new workload is invisible as overwork because it was absorbed voluntarily, one prompt at a time. Nobody forced the product manager to write code. The AI just made it possible, and the organization made it expected.

The ratchet never turns back. No company has introduced AI tools, documented the resulting productivity increase, and then reduced working hours accordingly. The efficiency gains become the new floor. The ceiling rises.

Robert Solow observed this pattern in 1987, before the internet existed: “You can see the computer age everywhere but in the productivity statistics.” Four decades later, the pattern holds. You can see AI everywhere but in the working hours.

§What the Intensity Paradox means

The promise of AI at work was simple: do more in less time, and keep the time. The Intensity Paradox reveals what actually happens: do more in the same time, then do even more, then burn out.

The Berkeley researchers recommend what they call “AI practice” — intentional norms around AI use, structured pauses before major decisions, sequencing work to reduce context-switching, protecting time for human connection. These are reasonable suggestions that will not be implemented at scale, because they require organizations to voluntarily limit the productivity gains that AI adoption provides.

The Intensity Paradox persists because the people who benefit from it — executives, shareholders, the 38% of C-suite leaders who report low burnout — are not the people who pay for it. The cost is borne by the associates and entry-level workers who adopted the tools they were told to adopt, took on the work they were enabled to do, and discovered that the reward for working at machine speed is the expectation of working at machine speed forever.

The technology that was supposed to give you your time back didn’t give it back. It gave you more work to fill it with, and then it measured how fast you could fill it.

Originally published at https://noahaust2.github.io/strategist-dashboard/blog/the-intensity-paradox.html