I recently attended a talk by Marshall Brain. He made the argument that the shift toward intelligent robots in the labor force will result in a much higher degree of structural unemployment in our economy. Brain presented some economic data which showed that gains in productivity over the past 10 years have not yield median wage gains or maintained a demand for labor. There were some details where I thought Brain undercut his plausibility (e.g., he didn’t seem to be using the most wildly accepted definition of a recession, instead simply focusing on the metric of full employment as the only worthwhile economic indicator). An irritated audience member asked him if he had heard about a discipline called history which suggests that these sorts of predictions never pan out.
Brain’s response was that the difference between labor saving technology in the past and the emergence of intelligent robots is that the latter are a new species. In other words, the demand for labor may increase with an expanding economy, but intelligent robots will have incredible comparative advantages compared to humans for the new occupational opportunities. Additionally, while technology expands and leverages human abilities, opening up opportunities, intelligent robots with agency would not need the complement of particular human cognitive skills.
The empirical argument is based on the peculiarities of the recent past, so the historical argument looking back to the “industrial revolution” and its effect on economic growth and long term demand for labor and the impact on wages is compelling. In Farewell to Alms Greg Clark points out that in fact since the Great Divergence the rise of the mass consumer society has been enabled by enormous comparative and absolute gains of wealth accrued toward unskilled labor. Only since 1970 has this dynamic been somewhat reversed. But I think that one point to remember is that the pattern of 1800-1970 itself is a relatively short time window. Obviously a future where intelligent robots substitute for human labor and management is not analogous to a pre-modern agrarian economy caught in the Malthusian trap, but I think it is important to remember that refuting Brain’s argument by an appeal to history itself relies on fixing a particular set of background conditions which themselves are relatively new.
All that being said, I can imagine intelligent robots replacing humans in a wide range of service, manufacturing and even professional jobs. Imagine for example a robot doctor who has immediate access to the total body of the latest research, but can intelligently weight these results appropriately so the swell of most recent likely false positives don’t hold so much sway. There would be no worries that a robot doctor would not be able to engage in Bayesian logic. But does that mean there won’t be roles for humans? Perhaps there are niches for human art and cultural production where robots might crave organic authenticity? After all, why couldn’t there be Bobo robots who are willing to shell out extra for human-made products which exhibit the imprecise, irrational and wild creativity characteristic of the organic substrate mind? Humans would have an enormous advantage over robots in Outsider Art. One could spin many scenarios of this sort, as Brain’s model seems to be predicated on the standard opposition between man and machine which goes back to the Luddites.
Finally, I wonder if a nation like Japan might not be well positioned if the rise of the machines does result in reduced demand for human labor. Japan’s native labor force is shrinking. This sort of thing is generally considered to be bad, but if robots entered the labor force and increased total productivity greatly then the problem of an imbalance between a large retired class and a smaller labor force goes away. The remaining younger humans in the labor force could focus on jobs where humans have a shot; e.g., instead of the Salaryman the Freeter might be the modal human Japanese.