A new technology promises to speed up the annoying parts of your job. Everyone gets excited about freeing up time for deep work and leisure. Then you end up busier than before without producing more of the high-value output that actually moves the needle.
This cycle has repeated with every major workplace technology - email, mobile computing, the entire front-office IT revolution of the 1990s. And it's playing out again with AI tools in 2026.
The Jevons Paradox, Applied to Your Calendar
The pattern has a name in economics: the Jevons paradox. When a resource becomes cheaper or more efficient to use, total consumption of that resource goes up, not down. Coal-efficient steam engines didn't reduce coal consumption in Victorian England - they made coal so useful that demand exploded.
AI writing tools follow the same logic. When it takes 10 minutes to draft an email instead of 30, you don't get 20 minutes back. Instead, you send three emails where you used to send one. When AI can generate a first draft of a report in seconds, the expectation isn't that you finish early - it's that you produce more reports, or that each report goes through more revision cycles, or that new reports get invented that nobody asked for before.
The time savings are real at the task level. But they evaporate at the job level because the freed-up capacity gets immediately filled with more tasks.
Who Captures the Gains?
There's a sharper version of this problem. When AI makes individual tasks faster, the productivity gains often flow to whoever sets the workload - managers, clients, the market - rather than to the person doing the work.
A freelance writer who can produce articles twice as fast doesn't work half as many hours. They take on twice as many clients, or their per-article rates drop because the market adjusts to the new speed baseline. A developer using Copilot or Cursor doesn't leave at 3pm. They're expected to ship more features per sprint.
This isn't unique to AI. It happened with spreadsheets (accountants didn't get shorter workweeks - they got more complex analyses to run), with email (response time expectations compressed from days to hours), and with smartphones (the entire concept of being "off the clock" dissolved).
So Is AI Actually Useless for Productivity?
No. But the gains are real only when you deliberately protect them. The people who genuinely work less or produce higher-quality output with AI tools tend to share a few traits:
- They set boundaries before adopting the tool. If AI saves you 5 hours a week, you decide in advance what those 5 hours are for - and you defend that time against new obligations.
- They use AI to raise quality, not just speed. Instead of writing three mediocre drafts, they write one and spend the saved time on research, editing, or thinking. The output gets better rather than just more voluminous.
- They resist the ratchet. When a manager sees faster output and increases expectations, they push back with data about quality vs. quantity tradeoffs.
The uncomfortable truth is that AI productivity tools work exactly as advertised at the individual task level. The problem isn't the tools. It's that most organizations and markets treat any efficiency gain as an invitation to increase throughput rather than reduce effort. Until that default changes - and there's no sign it will - "AI made my job easier" will remain the exception, not the rule.