Solow's productivity paradox
This is documented history rather than a laboratory result. The events are well recorded. What they teach is an argument, and the entry makes it openly.
By the 1980s, companies had spent enormous sums putting computers on desks across the economy, convinced this would transform how much work got done. But the productivity statistics stubbornly refused to show it. In 1987, the Nobel-winning economist Robert Solow captured the puzzle in a single famous remark: you could see the computer age everywhere except in the productivity statistics. It became known as the productivity paradox. The eventual lesson was uncomfortable. A powerful new tool simply runs whatever process already exists, faster, rather than fixing a broken or badly designed one. Bolt fast computers onto tangled, poorly designed workflows, and you get the same tangled results more quickly, plus a large bill. The gains came only later, and only where organisations redesigned how they actually worked, rather than just buying the technology.
A powerful tool accelerates whatever process it is applied to, which means it makes a good process better and a broken process worse, faster. This is the trap waiting for every organisation rushing to adopt the latest technology, today most obviously artificial intelligence: bolt it onto a broken workflow and you get the same bad decisions at higher speed, plus the cost of the tool. The spending shows up on the invoice, not in the results. Before you add a powerful new tool, fix the underlying process, because the tool will faithfully amplify whatever it finds, and if what it finds is a mess, it will produce a faster mess.
The same question returns with every wave of new technology, and it is being asked about AI now, usually in the form of a question about return on investment. A company buys the tools, individual tasks visibly speed up, and the quarterly numbers look much as they did before. Solow's observation explains why. Speed lands on whatever the company was already doing, so output moves only where the work itself was redesigned around the tool. Where the workflow was left alone, the company has bought faster versions of its existing habits.
And the time a good tool does save rarely stays saved, which is the Jevons paradox.
Solow says a tool accelerates whatever process it lands on, so where the work was not redesigned around it the company has bought faster versions of its existing habits and the productivity numbers do not move. Jevons says that where a tool does save time, the saving gets spent rather than banked, because making something cheap creates more demand for it. They look like the same complaint about technology and they are two questions in sequence. First, did the tool save anything, which is Solow. Second, where did the saving go, which is Jevons. A company can fail either one. The one that passes Solow by redesigning the work then meets Jevons, and the only defence against the second is to decide in advance what the freed time is for.
Source: Solow, 1987. The subsequent literature is known as the productivity paradox.
Brynjolfsson and McAfee, 2014
On why technology's gains arrive only when the work around it is redesigned, not just when it is bought.
Draw your own card. It does not take long, and it rewards taking your time.