Every time I talk about autonomous robotaxis ending driving jobs, someone hits me with the same analogy.
When ATMs rolled out, everyone predicted the death of the bank teller. Instead, teller jobs grew for decades. ATMs made branches cheaper to run, so banks opened more branches, and tellers shifted from counting cash to selling products.
The conclusion people draw: automation doesn't kill jobs, it changes them. Unfortunately, the ATM is the wrong analogy for robotaxis, and here's the test that proves it.
Test 1: A part, or the whole thing
Is the human a task inside the job, or is the human the entire cost of the service?
A teller did 20 things back in the day. The ATM automated some of those tasks, but not all of them.
Now take Uber. A driver does one thing. Drive. The driver is the single largest cost in every ride you've ever paid for. When you automate the driver, there's no remaining job to shift into. It's not like the "driver" will stick around so you'll continue to have someone to speak to or adjust the temp.
When automation removes a part of the job, the human adapts. When automation removes the entire job, the human is no longer needed.
Test 2: Follow the price, not the technology
Ask one question about any automation: did it reduce the price the consumer pays?
The ATM didn't. In the decades after ATMs rolled out, your checking account didn't get 10X cheaper. Banking fees have only gone up since, not down. The ATM automated one task inside a bank, and the savings stayed inside the bank.
Now run the test on driving.
An Uber runs you roughly $2/mile. An autonomous vehicle, with no driver to pay, running near 24/7 utilization, eventually gets that to around $0.30/mile.
That's not a task getting automated inside a business. That's the price of moving a human body through space dropping 85%. The ATM never did anything like that.
The right analogies
If you want historical parallels for autonomous robotaxis, here are the two that actually match.
Switchboard operators. In 1950, about 350,000 women worked as operators for the phone company, with hundreds of thousands more running switchboards in offices and hotels. Roughly 1 in 13 working American women was a switchboard operator. The operator was the cost of connecting a call. Automated switching removed her, calls got dramatically cheaper, call volume exploded, and the job went to zero. More calls than ever. No operators. That's the pattern: demand for the service goes up, demand for the human goes away, because the human was the price.
Elevator operators. For decades, every elevator had a human driving it. Automatic elevators existed as early as 1900, but riders didn't trust them, the same way people say they'll never get in a car with no steering wheel. Then a 1945 operator strike shut down New York City, building owners got serious about automation, the industry spent a decade building public trust, and the job vanished. Economist James Bessen reviewed all 270 occupations in the 1950 census and found exactly one that automation fully eliminated: elevator operator. A human whose entire job was operating a vehicle, kept around mainly because passengers didn't trust the machine.
Sound familiar?
The rule
Next time someone hands you an automation analogy, run the two questions:
Does it replace the entire job, or a task inside it?
Does it collapse the price to the consumer?
The ATM fails both. That's why tellers survived. Autonomous robotaxis pass both. The human driving is the entire job, and removing the driver collapses the price of a mile traveled by 85% for consumers.
This means ride-share driving is most likely to go the way of the switchboard operator and the elevator operator. Jobs that no longer exist.