Despite all the dazzling digital advances, the trillions of dollars spent on computer technology have done almost nothing to make the world a more productive place. The economist Robert Solow, who identified this problem, called it the productivity paradox. In 1987, a decade into the computer revolution, he observed that productivity growth had actually slowed down. “You can see the computer age everywhere,” he wrote, “but in the productivity statistics.”
Economists and historians have spent a lot of time scratching their heads about why this might be. But you already know the answer: Software is supposed to make us faster, but often it makes us slower. We spend a half-hour clumsily filling out PDF files that we could have done with pencil and paper in a minute. We spend an hour bouncing emails back and forth to clarify a point that could have been nailed down in 30 seconds on the phone. The digital era has made a lot of everyday work more complicated and less efficient than it was 30 years ago.
The huge productivity gains of the industrial age didn’t happen just because someone invented a new technology; they happened because people also figured out how best to reorganize work around that technology. A steam engine, for example, would have been no use to textile manufacturing if textile workers had remained a scattered network of independent farmers, as opposed to a group of employees gathered under a single factory roof. And conveyor belts were nothing new when Henry Ford put them to use in his factory; the revolution was how he arranged for workers to use them, breaking the complex work of automobile manufacture into repetitive, specific tasks. Ford’s breakthrough was as much organizational as technological.
Computers have failed to produce a huge surge in productivity, but the problem isn’t the computers. It’s that we haven’t let workers tap into the computers’ true power — automation. We still use them like typewriters or calculators.
The arrival of ChatGPT — most of all, its remarkable ability to write computer code to automate well-defined tasks — can change all that. Instead of eliminating many white-collar jobs altogether, as people are understandably worried it will do, it has the ability to do something much more powerful: to eliminate what’s boring about those jobs, freeing us up to be more stimulated, more creative and more human in our work. In the process it can drastically increase productivity.
Most office jobs today are a matter of manipulating data. There are a lot of things A.I. can’t do, but it’s very good at writing code to manipulate data. Office workers all just got their own personal tech consultant. They just need to learn to use it.
As a historian, I confess that I was quick to sneer at the idea that ChatGPT could ever do any part of my job. I mean, have you ever asked it to interpret the causes of World War I? It gives you a listicle of contributing factors. And the writing — don’t get me started on the writing.
But then I got the idea to ask ChatGPT to write computer code to analyze data sets, which is laborious work that, as an economic historian, I have to do often. There are things I know how to code, and there are things I wish I knew how to code. ChatGPT could do both, easily. Boring, repetitive tasks that I knew a computer should be able to do, but that I didn’t know how to make it do, suddenly became as easy as typing in my request.
If a historian can do it, anybody can.
ChatGPT might not be able to help someone at a medical supply company figure out why and where a particular batch of medicine was misplaced — that might require real ingenuity. But it could take over the tedious business of tracking orders and deliveries on an Excel spreadsheet, creating more time for the clerk to do precisely that kind of more challenging and satisfying problem-solving work.
Researchers at the Massachusetts Institute of Technology recently conducted an experiment, reported in a working paper, with 444 “college-educated professionals” who were given a “midlevel professional writing task” like drafting news releases or delicate emails. Half were provided with ChatGPT and half weren’t. The participants who were given ChatGPT took less time, wrote better and reported enjoying the task more. Even more important, perhaps, ChatGPT helped “low-ability workers,” meaning that those with weaker writing skills — but perhaps with good ideas — could carry out the task effectively.
ChatGPT can do that for writing assignments, but it can do it even better for coding tasks. Seemingly impossible undertakings, like making a new dashboard to track heat maps of weekly sales in Excel, will be easy. By breaking down complex analytic problems into small steps, like Ford engineers did for the Model T, employees will be able to make their own assembly lines of data, freeing them to do more creative work. Call it everyday automation.
I realize that automating your everyday tasks can be scary. If a macro can generate your daily report in five seconds rather than the five hours it takes you, what is your value? It’s tempting to see ourselves, or our employees, as nothing more than those repetitive tasks. We can’t imagine a world where those same employees could do more.
But if a company can take this momentous step from avoiding to embracing everyday automation, it will have a competitive edge. Companies that promote workers who can automate the tedious parts of their jobs will be more profitable in the long run, because those employees can then do more complicated, more rewarding, more human work. Almost by definition, the work that cannot be automated will be better paid.
Until now, you’d have to ask the I.T. department to help automate part of your work flow. But with ChatGPT anyone will be able to do it, with just a little training. As with Ford’s assembly line, the challenge today is no longer technological but organizational.
It’s true that some shortsighted corporations would be happy to do the same work they do now but with fewer people. But my suspicion is that most successful businesses will realize the long-term potential of encouraging workers to solve harder problems.
Changing the way companies are organized is much harder than upgrading software. Despite what you read in the news, most of us don’t work at Google or Amazon. We work at the same sorts of companies that existed in 1973, that tried to go paperless but never quite made it. Real change might take a generation or two — but hopefully not.
Everyday automation, if it happens, will be the undoing of Henry Ford. His assembly-line production paid workers better but was dehumanizing. It implied that the only way we could be more productive and make more money was to become more like machines. Everyday automation says the opposite: that the way to be more productive and earn more money is to use our technology to become more human again.
Louis Hyman is a historian at the I.L.R. School at Cornell and the author, most recently, of “Temp: How American Work, American Business and the American Dream Became Temporary.”
This article originally appeared in The New York Times.