Who has the most to fear from greater adoption of robots in the workplace? Historically the emergence of new general purpose technologies has been associated with an increase
in employment, but many people worry that AI and robotics might have the opposite effect. The most comprehensive review I’ve seen, this long but excellent piece
by Scott Alexander, suggests that, at minimum, technological unemployment isn’t happening yet
(though Alexander’s overall conclusions are quite pessimistic).
We’re a long way off human redundancy (I hope), but it’s interesting to look at the employment impact of the growing number
of industrial robotics installations. A new(ish) study
by Jay Dixon, Bryan Hong and Lynn Wu has an interesting finding: robotics investments are associated, at the company level, with increased employment overall, but decreased
employment among middle managers. Why? Because robot-led processes have much less variance than human-led ones and so require less management.
This is the opposite of many people’s intuition that low-skilled work is the first to be automated away (though there is a big literature on the displacement of middle-skilled work - see, e.g., here
). That said, I’m skeptical of general optimism about the sustainability of automation-complementary labour. As I noted in TiB 136
, in chess there was a brief period of excitement about human-machine collaboration before machines’ dominance became total
. It’s early days for robotics.