AI is going to kill all the jobs - isn't that wonderful?

An interesting little bit of research about an entire sector of jobs that really were killed off by automation - telephone operators.

Telephone operation was among the most common jobs for young American women in the early

1900s. Between 1920 and 1940, AT&T adopted mechanical switching technology in over half of

the U.S. telephone network, replacing manual operation. Although automation eliminated most of

these jobs, it did not affect future cohorts’ overall employment: the decline in operators was

counteracted by reinstating demand in middle-skill clerical jobs and lower-skill service jobs.

Using a new genealogy-based census-linking method, we show that incumbent telephone

operators were most impacted, and a decade later more likely to be in lower-paying occupations

or have left the labor force

AI taking all the jobs (or more likely, many in some sectors, few in others) is not a problem over time. Mechanisation works just like us free marketeers say it does over those long periods of time. Simply because the young grow up and do other things other than that which has now been automated.

Where there is a potential problem is in those thoroughly trained in that old thing now gone and don’t or can’t retrain across to those other, newer, things that are now to be done.

Another way to put this is that it is the speed of transition that matters. A transition that takes place over a working lifetime doesn’t matter at all. One that happens tomorrow might well do.

But that’s not the end of the story. There always is jobs churn in the economy anyway. And it’s much, much, higher than people generally think it is. A rough guide is that 10% of all jobs in the economy are killed off each year, another 10% newly created each year. There’s near always some technological shift between the old and the new as well. Unemployment is not this flow from job to job, the unemployment number we see quoted is those who get stuck in this position of no job, not those moving across from old to new.

At which point, if the job destruction/creation rate, as a result of the new technology, rises substantially above that normal societal rate then perhaps we might actually have that claimed problem of technological unemployment. If not we won’t.

Currently the management consultancy predictions are of the order that AI endangers 40% of jobs over the next decade or two. A period of time in which we expect 100 to 200% of all jobs to be destroyed anyway. This is not to say that there’s going to be no problem here - there are always problems with humans. Rather, the jobs inferno about to be brought about by AI seems well within the usual limits of the jobs inferno that always does exist within the economy. We might even have marginal problems but we’ve not got a large and systemic one.

Another way to put this is that we’ve already got the systems in place to deal with the problem - a free market in labour and an unemployment system for those who get stuck, temporarily, in the transition. What else do we need?