The AI Paradox: 4 Reasons Why Making Work Cheaper Creates More of It
History shows efficiency doesn't reduce work — it explodes it. Why AI will create more opportunities, not fewer, through the lens of Jevons Paradox.
Andrej Karpathy put the current anxiety plainly:
A magnitude 9 earthquake is rocking the profession... powerful alien tools handed to us without a manual. [2]
But the historical pattern points somewhere unexpected.
Major efficiency gains rarely lead to less work. They tend to produce an explosion of it.
1. Jevons Paradox Comes for Knowledge Work
The Core Insight
When 19th-century technology made coal more efficient, total coal demand didn't decrease—it skyrocketed. Making a resource cheaper unlocks use cases that were previously unimaginable.
This pattern has repeated in computing:
| Era | Units |
|---|---|
| Mainframes | Hundreds |
| Minicomputers | Tens of thousands |
| PCs | Millions |
Each leap: roughly 100x growth.
Now it's happening to knowledge work: reviewing contracts, writing code, generating campaigns, market research, customer support.
Jevons paradox is coming to knowledge work. By making it far cheaper to take on any type of task, we're ultimately going to be doing far more.
— Aaron Levie, CEO of Box [1]
2. Your "Boring" Engineering Skills Are Now a Superpower
The assumption is that foundational skills are becoming obsolete. The data says otherwise.
Google Cloud's Addy Osmani points to the skills that separate teams succeeding with AI from teams struggling:
- CI/CD pipelines
- Testing
- Documentation
- Code review
These foundations are what turn AI agents from chaos generators into actual productivity tools.
Developers who already invested in CI/CD, testing, documentation, and code review are having the most success with AI tools. These "boring" foundations are accelerators.
— Addy Osmani [4]
3. The Job Shifts from 'How' to 'What'
The nature of the work is shifting. Less implementation, more direction.
The New Reality
The bottleneck is shifting from "can we build this?" to "should we build this?"
Boris Cherny, creator of Claude Code, offers a stunning example:
In a single month, he landed 259 pull requests—over 78,000 lines of code changed—with every single line written by AI.
Increasingly, code is no longer the bottleneck.
— Boris Cherny [3]
The human role has evolved into that of a director, guiding powerful tools to achieve a strategic vision.
4. Fortune 500 resources in every entrepreneur's pocket
Historically, large corporations could absorb high investment costs. Small teams faced brutal tradeoffs: marketing vs. product vs. support?
AI removes that constraint.
Imagine a 10-person services firm that never had custom software because the project was too large to start. Now someone on that team can build a prototype in days.
AI gives every business access to the talent and resources of a Fortune 500 company 10 years ago.
— Aaron Levie
The pattern
It's happened before:
| Era | What Got Cheaper | Result |
|---|---|---|
| 1800s | Coal efficiency | More coal consumption |
| 1990s–2020s | Computing | More software everywhere |
| 2025+ | Knowledge work | More projects, more experiments, more innovation |
We don't do less work when things get cheaper. We discover there was more worth doing than we had the capacity for.
The Critical Question
Are we prepared for a world where the bottleneck shifts from "can we build this?" to "should we build this?"
Key Takeaways for Your Career
- Invest in judgment, taste, and strategic thinking—these become the differentiators
- Double down on "boring" foundations—testing, CI/CD, documentation
- Think like a director—guide AI tools rather than compete with them
- Lower activation energy—projects once too expensive to start are now viable
The earthquake is real. But it's creating new ground, not just swallowing the old.
References
[1]: Aaron Levie (CEO @ Box) — tweet
[2]: Andrej Karpathy (AI researcher, founder of Eureka Labs) — tweet