If you look at a graph of productivity for the last 45 years, it would look like a zig-zag sideways pattern ranging between zero and 5% with few exceptions, most relating to brief spikes at the end of recessions when output rises faster than employment for obvious reasons. The PC came on the scene in the early 1980s growing in importance for most of the next two decades. One might think of the PC as the ultimate engine of productivity improvement. But it wasn’t. Neither was the advent of the Internet in the mid-1990s. Robots replaced workers in manufacturing plants but the output per man hour didn’t change significantly. Now we enter the AI era. A computer will replace the order taker soon. Call centers will be manned by non-humans as well. There is no doubt that technology displaces workers. But does it make workers more productive? One would think so.
Look at retailing. Zombie malls reflect hundreds of store closings as we all enjoy the convenience of shopping from our living rooms. Store salespeople and check-out clerks are displaced. But drive down the highway and look at all the distribution centers replacing those zombie malls. 1,000 yellow cabs in New York have been replaced by 5,000 Uber drivers. Look at the masses of Amazon and UPS trucks delivering to you what you would otherwise buy in the store. Don’t mistake convenience with enhanced productivity.
Don’t misunderstand me. What I am trying to say, is for all the decades of promises I have listened to in this economic world we live in, I try to look at the reality, not the hype. There is no doubt AI will change the way we live just as PCs, streaming, smartphones, and robotics have done in the past. But do we produce more today? Can we attribute the economic revenue we produce directly to the technological tools at our disposal? My accountant could produce more tax returns because the computer helps to enter data more easily. But the gross number of returns produced in this country relates to population, not technology. And if TurboTax allows more of us to do our own returns, what then?
Technology displaces workers. To the extent the remaining workers can produce more with less, it enhances productivity. That broadens the pie and allows economic growth. But what happens to the displaced workers? If they become Uber drivers or Amazon delivery people, is there as big a change in overall productivity as one expects?
Last week, Doug McMillon, CEO of Wal-Mart said prophetically that in 5 years, when Wal-Mart is roughly a third bigger than it is today, it will have the same sized work force that it has today. Wal-Mart will be more productive. McDonald’s can probably make the same statement. Others, not so well managed, will still be able to use AI features to increase productivity. But that doesn’t mean national productivity will increase at the same pace. Granted, companies that don’t keep pace will be left behind. Look at the Zombie malls I mentioned earlier. Right now, most of our Mag 7 companies are spending hundreds of billions of dollars on chips, data centers, power plants, etc. to be able to help companies become more productive. To date, the revenues produced don’t equal the cost. Over time, I suspect they will. Without getting too technical, the chips designed for the PC revolution, microprocessors that worked serially in computation, are being replaced by GPUs (graphical processing units) that by working in parallel are much more powerful. This transformation is still in the early stages. Undoubtedly, I will get information faster. It may make me smarter, but will it increase national productivity?
Industrial revolutions in the past improved productivity. Cars, then trucks moved goods faster. The telephone replaced the telegraph. Planes allow movement of people and goods with much greater efficiency. Won’t AI do the same? The jury is still out. Look at social networking. We almost all do it. It changes the way we communicate (not always for the best). But does it move the productivity needle? It certainly has created lots of jobs for Facebook, TikTok, and X. But I doubt whether the billions spent has done much to increase production per man hour. Some AI dollars will enhance productivity. Some will be misdirected. Harmful uses may subtract from productivity gains.
Why all this discussion? Because one measure of GDP growth is population growth time productivity improvement. We know population growth is going in the wrong direction. Births are down, death rates are rising, ex the impact of Covid, largely due to aging, not only in the US but also around the world. Immigration has dropped sharply. Population shifts just in the last two years, have knocked about 0.3-0.4 percentage points off of GDP. That doesn’t sound like a lot but if the national goal is to take normalized growth from 2-3% to 4-5%, it will require a massive and sustained boost in productivity. That may or may not happen. But the average gain for 2025 to date is 1.2%. If AI is the productivity kicker of the future, its impact hasn’t been felt yet. Also worth remembering that while those closest to AI and its improvements will see productivity improvement in the coming years, productivity includes those Uber drivers, construction laborers, housekeepers and small business operators.
None of this suggests we aren’t growing. Population is still rising and productivity gains remain within the range of the last 50 years. There is no reason to expect growth to stop or even slow although lower immigration and tariffs might be a temporary headwind. Maybe AI will live up to its hype and expand productivity in ways not seen for more than a century. But let’s not get too excited by the hype until we see some signs that technology is making the kind of impact the invention of trains, planes and automobiles made more than a century ago.
Today, Kevin Durant is 37. Andrew Dice Clay turns 68.
James M. Meyer, CFA 610-260-2220