If AI Is Making Us More Productive, Why Isn't GDP Reflecting It?
If AI Is Making Us More Productive, Why Isn't GDP Reflecting It?
Everyone is using AI. Productivity should be through the roof. But GDP growth looks normal, and most companies report that AI has not meaningfully changed their output. What is going on?
The answer is that most AI usage is not actually productive. It is busywork that feels productive.
The Busywork Trap
Look at what most people use AI for:
- Rewriting emails to sound more professional (the original was fine)
- Generating reports that nobody reads
- Creating meeting summaries that sit in a folder
- Making slides prettier instead of more informative
- Drafting social media posts that get 12 impressions
None of this creates measurable value. It is the digital equivalent of reorganizing your desk - it feels like work but nothing actually changes.
What Real Productivity Looks Like
Real productivity gains come from eliminating work entirely, not from making busywork faster. When a desktop agent fills out 50 CRM entries that would have taken two hours of manual data entry, those two hours are genuinely freed up. When it auto-organizes 200 files into the right folders, that is real time saved.
The difference is measurable. You can count the hours saved. You can point to tasks that simply do not exist anymore because an agent handles them automatically.
Why the Gap Exists
The gap between "everyone uses AI" and "GDP is flat" exists because most AI usage optimizes for appearance rather than output. A polished email does not close more deals. A prettier report does not change the decision it informs.
The companies that actually see productivity gains are the ones automating the mechanical work - data entry, file management, repetitive browser tasks, form filling. These are boring, invisible improvements. Nobody posts on LinkedIn about their CRM being automatically updated. But the cumulative time savings are real and they show up in the numbers.
The question is not whether AI can make us more productive. It is whether we use it for real work or just fancier busywork.
Fazm is an open source macOS AI agent. Open source on GitHub.