I saw someone who was worried that AI was gonna cause real economic trouble soon by replacing travel agents. But the advent of the internet made travel agents completely unnecessary, and it still only wiped out half the travel agent jobs. The number of travel agents has stayed roughly the same since 2008!
This reminds me of Patrick McKenzie’s tweet thread:
Technology-driven widespread unemployment (“the robots will take all the jobs”) is, like wizards who fly spaceships, a fun premise for science fiction but difficult to find examples for in economic history.
(The best example I know is for horses.)
I understand people who are extremely concerned by it, but I think you need a theory for it which successfully predicts bank teller employment trends over the last 50 years prior to justifying the number of column inches it gets.
“Haha Patrick you’re not going to tell me bank teller employment is up since we made a machine literally called Automated Teller Machine are you.”
Bessen, 2015:
“Wait why did that happen?”
Short version: the ATM makes each bank branch need less tellers to operate at a desired level of service, but the growth of the economy (and the tech-driven decline in cost of bank branch OpEx) caused branch count to outgrow decline in tellers/branch.
The obvious response, which I thought of as soon as I saw this, is indeed contained in multiplereply tweets:
How about bank teller employment divided by total population size?
Ok but USA population went up +20% from 1990 to 2010 so tellers per capita did decrease over this period.
(McKenzie did not reply to either of these, for some reason.)
If you don’t normalize for population, graphs and claims like this are profoundly misleading. (Similarly to normalizing geographic data for population density, correcting for inflation, etc.)
This reminds me of Patrick McKenzie’s tweet thread:
The obvious response, which I thought of as soon as I saw this, is indeed contained in multiple reply tweets:
(McKenzie did not reply to either of these, for some reason.)
If you don’t normalize for population, graphs and claims like this are profoundly misleading. (Similarly to normalizing geographic data for population density, correcting for inflation, etc.)