Companies trying to "replace" people with artificial intelligence fall into a trap described by one of the most important economists of the digital economy. Instead of increasing productivity, they reduce the value of human labor — and their own competitive advantage.
Why Isn't AI Delivering Productivity Yet?
Erik Brynjolfsson — Stanford professor, director of the Stanford Digital Economy Lab, and one of the most influential economists studying digital transformation — has spent years researching why general-purpose technologies don't produce immediate productivity gains. His answer is simple, though uncomfortable for boards seeking quick returns: investments in AI — in new competencies, processes, and organizational changes — are invisible in traditional productivity measures. Brynjolfsson calls this the J-curve of productivity.
What Does the J-Curve Mean?
Companies invest — often heavily — but the return comes with a delay. It was the same with the electrification of factories, which took decades before delivering its full benefits. From my observations of the Polish market, many boards interpret this natural "dip" of the J-curve as a failed implementation — and give up too early. Brynjolfsson argues that the J-curve for AI will be steep and faster than for electricity, but measured in years — not quarters.
The Turing Trap
Brynjolfsson's most important warning is what he calls the "Turing Trap" — optimizing AI to perfectly imitate humans rather than augment them. The logic is merciless: if a machine perfectly replaces a human, the value of human labor drops. Companies that prioritize automation over augmentation don't just lose talent — they reduce their own capacity to create value that a machine cannot produce on its own.
My Conclusion for Boards
Don't ask: "How many positions can we replace with AI?" Ask instead: "In which processes will AI strengthen people to do things they couldn't do without it?" Because the data — both Brynjolfsson's and the reports I've been analyzing for years — consistently show one thing: productivity grows fastest where humans and machines collaborate. Not where one replaces the other.
If you believe this topic is relevant to your company and you'd like to discuss how to effectively align AI solutions with your company's reality, I invite you to get in touch — Leszek Giza.