The NYT gave us yet another piece telling us that Donald Trump is right about his growth projections and that the Congressional Budget Office is wrong. The piece, by Kai-Fu Lee, the chairman and chief executive of Sinovation Ventures, a venture capital firm, and the president of its Artificial Intelligence Institute, tells readers that we are about to see mass displacement of jobs due to the spread of artificial intelligence (AI).
June 25th, 2017 [CEPR]
This mass displacement has another name, it’s called “productivity growth.” In other words, Lee is predicting a massive boom in productivity growth. If we get a massive boom in productivity growth, it will mean a huge rise in the rate of GDP growth.
While Lee doesn’t put a number on the rate of productivity growth, it is clear he thinks it is faster than anything we have seen in the past. In the long post-war Golden Age from 1947 to 1973, and again from 1995 to 2005, productivity growth averaged 3.0 percent annually. (This was a period of rapid wage growth and low unemployment.) Since Lee apparently thinks the growth will be even faster with his job-killing AI story, we should probably envision productivity growth even faster than this 3.0 percent rate.
In that case, Trump and his crew are probably being too pessimistic projecting GDP growth of just 3.0 percent over the next decade. After all, GDP growth is just the sum of productivity growth and labor force growth. Even with the retirement of the baby boomers we are still expecting labor force growth in the range of 0.5–0.7 percent annually. So, if Mr. Lee is anywhere close to being right about his projections of the future, then the Trump team is being too pessimistic.
We can leave the resolution of this debate over the future for other occasions, but there is one point that is clear. If anyone thinks that Mr. Lee’s view should be treated seriously, they better also take Trump’s growth projections seriously. Anyone who thinks this NYT column is plausible but that Trump is just inventing numbers has problems with simple arithmetic and should be laughed out of any serious policy discussion.
There is another important point that Lee misses in his column. He argues that AI will transfer wealth from the rest of us to the people who own AI. This is sloppy thinking. One gets to “own” AI from patent and/or copyright monopolies. These come from governments, not technology. If the ownership of AI is leading to an upward redistribution of income the most obvious way to deal with it is to reduce the length and strength of these monopolies.
This basic point, that policies designed to give incentives to innovate can be altered should be obvious to anyone involved in this debate. But, as we all know, the economy is suffering from a severe skills shortage.
