The Efficiency Trap: Why AI Might Be Making You Work More, Not Less

A few months ago, I caught myself doing something I thought AI was supposed to prevent.

I had just shipped a feature for Modulyo in a fraction of the time it would have taken a year ago. Claude handled the code. I handled the direction. The loop closed faster than it ever had. And instead of stepping away from my desk, I opened a new task. Then another. By Friday, I hadn't reclaimed any time. I had simply filled the space AI created with more work.

I didn't recognise what was happening until I came across Cal Newport's framing of a 19th-century observation. The Jevons Paradox.

What is the Jevons Paradox, and why does it apply to AI-assisted work?

The paradox is simple: efficiency gains tend to increase total consumption, not reduce it. When steam engines became more efficient in the 1860s, coal usage went up, not down. More efficient engines meant cheaper energy, and cheaper energy meant more demand for everything that ran on it.

Newport's insight is that knowledge work follows the same logic. If AI makes you more productive, your services effectively get cheaper. Cheaper services attract more demand. More demand means more work. The net result isn't a shorter workday. It's a fuller one.

Why does getting more efficient with AI lead to higher demand, not more freedom?

The mechanism works at two levels. Externally: when you can deliver more, faster, clients expect more, faster. The market recalibrates around your new capacity. Pricing follows capability, and capability no longer has the same friction it once did. The ceiling rises. So do the expectations.

Internally, it's quieter but just as real. If you're driven, curious, or building something you genuinely care about, efficiency doesn't create rest. It creates room for the next idea. The next feature. The next piece of content. There's always something worth doing with a free hour.

This is the trap. Not malicious. Not obvious. Just the natural result of more capacity entering a system where the work, if you let it, expands to fill all available space.

How does this land differently when you're building alone?

In a company, efficiency gains get absorbed by the organisation. The increased output feeds a team, a product roadmap, a quarterly target. It gets diluted across dozens of people and priorities.

As a solo founder, there's no dilution. The gains arrive directly on your desk. And unlike a salaried employee, you also bear the personal cost of the decision: if you work more, you feel it. If you protect your time, you feel that too.

This is what makes the Jevons Paradox genuinely dangerous for solopreneurs. We're the most exposed. The efficiency gain and the lifestyle impact sit in the same person, with no buffer in between. I've watched this pattern in my own work and in conversations with other solo founders building on AI. The tools are extraordinary. The output is real. But the hours haven't shrunk. They've shifted.

What does using AI as a life-first founder actually require?

It requires a deliberate decision about what to do less of.

The efficiency AI provides is genuinely neutral. It can become more capacity for more work. Or it can become the same work done in less time, with the remaining hours protected for something else. The paradox doesn't resolve itself. You have to resolve it.

This connects directly to how I think about energy management. Just as I structure my day around cognitive peaks rather than calendar slots [link to energy scheduling post], I try to structure my AI use around protected outcomes rather than open capacity. The two questions aren't separate: when does your best thinking happen, and what are you actually trying to buy back with the time AI returns?

A life-first business isn't just a philosophy about values. It's an architectural choice you make every time a new tool arrives. You have to explicitly decide: am I using this to expand my output, or to shrink my hours?

For me, this means treating the time AI returns as a protected resource, not an open buffer. When a task takes two hours instead of six, the four hours don't automatically become four more hours of work. They become a longer walk with my daughter, a proper dinner, an hour of reading that has nothing to do with a product. Those hours have already been claimed. The AI didn't give them back by accident. I took them deliberately.

The Jevons Paradox is a warning, not a sentence. You can choose differently. Most people don't, because the pull toward more is strong and the case for less requires making peace with your own ambition.

That tension doesn't go away. I still feel it. But I've stopped pretending AI will resolve it on its own.