As I said in my last response, for well over a century we’ve seen a regular thirty-year cycle of booms and busts of hype, concern, and investment regarding automation. We are now near the peak of another boom. However, over this history the economic impact of automation has grown at a relatively steady rate, unrelated to these booms and busts. Thus the obvious question is: is this time different?
This is not to ask if there will be any tech progress or economic impact, nor if there’s been a recent increase in concern or investment, nor if some have recently claimed that this time is different. No, it is to ask if we have concrete evidence of substantial deviations from prior trends in automation impact.
Carl Frey says “there are already clear warnings”. In support, he (1) invokes the concept of a “general purpose technology”, (2) notes that artificial intelligence [AI] researchers are publishing more on application topics since 2009, (3) notes that since 2012 we’ve seen faster gains on AI research benchmarks (mainly due to to unsustainable increases in computing hardware budgets), and (4) notes that many experts predict rapid and broad change soon.
But these are all indicators suggesting that we are in a boom of hype, concern, and attention; they are NOT indicators that the economic impact of automation is deviating from long-term trends. We do NOT today see deviations from prior rates of change in growth, innovation, productivity, automation, or employment that we can clearly attribute to automation. As I said in my last response, the best predictors of which U.S. jobs are how automated are relatively mundane, they predict well, and they have not changed noticeably in two decades.
So yes, clearly we are in a boom of hype and concern. But no, we as yet have not seen evidence of a revolution in the patterns or quantity of automation. And since, over the last two decades, automation increases have on average predicted neither changes in pay nor employment, workers probably have little to fear even if such a trend-deviating revolution does come.
Many of you readers are young enough to live to see another such boom in your lifetime, peaking near roughly 2045-2050. If you do, remember this exchange and ask yourself then: Are the indicators that advocates use to justify concern then the needed indicators of actual and substantial trend-deviating economic impact, or are they instead only of recent hype, concern, and investment? If this next boom is like the last four, the answer will likely again be "no, this time is again not different."