The Paradox of AI Productivity: Why Your New “Partner” Might Be Burning You Out
AI was supposed to make work easier. Instead, it’s making work more.
That’s the sobering finding from a new Berkeley Haas study of 200 employees at a U.S. tech company, published in Harvard Business Review. And Simon Willison—one of the most thoughtful voices in AI development—confirms it matches his own experience: the productivity boost from LLMs is genuinely exhausting.
The Core Insight
The researchers identified a dangerous new pattern: AI introduces a rhythm where workers manage multiple active threads simultaneously. They write code while AI generates alternatives. They run multiple agents in parallel. They revive long-deferred tasks because AI can “handle them” in the background.
The feeling? Having a “partner” that enables momentum.
The reality? Continual attention-switching, constant output checking, and a growing pile of open tasks creating cognitive overload.
“While this sense of having a ‘partner’ enabled a feeling of momentum, the reality was a continual switching of attention, frequent checking of AI outputs, and a growing number of open tasks.”
This isn’t laziness or poor time management. It’s a structural problem with how AI augmentation actually works in practice.
Why This Matters
Willison’s personal account is striking: “I’m frequently finding myself with work on two or three projects running parallel. I can get so much done, but after just an hour or two my mental energy for the day feels almost entirely depleted.”
He’s also observing this in others: “I’ve had conversations with people recently who are losing sleep because they’re finding building yet another feature with ‘just one more prompt’ irresistible.”
This creates a dangerous organizational blind spot. Traditional productivity metrics look great—more output, faster delivery, more features shipped. But they’re masking unsustainable intensity that leads to burnout.
The study authors warn: it’s now “harder for organizations to distinguish genuine productivity gains from unsustainable intensity.”
Key Takeaways
AI is addictive. The “one more prompt” pattern mirrors other compulsive behaviors. Recognize it.
Parallelism has cognitive costs. Just because you can run three workstreams doesn’t mean you should.
Mental energy depletion is real. High output for two hours can leave you useless for the rest of the day.
Organizations need “AI practices.” Structured approaches to how AI is used, including limits and guardrails.
Existing intuitions are broken. Decades of knowledge about sustainable work practices no longer apply. We need new norms.
Looking Ahead
Willison puts it perfectly: “I think we’ve just disrupted decades of existing intuition about sustainable working practices. It’s going to take a while and some discipline to find a good new balance.”
The solution isn’t to stop using AI—the productivity gains are too valuable. But organizations and individuals need to develop new frameworks for sustainable AI-augmented work. This means:
- Setting explicit boundaries on parallel workstreams
- Building in recovery time after intensive AI sessions
- Tracking cognitive load, not just output
- Creating team norms around AI usage patterns
The irony is rich: the tool designed to reduce work requires us to build systems to protect ourselves from working too much. Welcome to the paradox of AI productivity.
Based on analysis of “AI Doesn’t Reduce Work—It Intensifies It” (HBR) via Simon Willison