December 01, 2013
Your Job Susceptible To Automation?
Back once again to the question of how fast jobs will get destroyed by automation as computers gain the ability to do tasks which previously only brains could do. A couple of Oxford academics think in the next 20 years machine learning and mobile robotics could even cut into a lot of jobs, including previously safe service jobs.
Certain job fields were cited as being at especially high risk for getting automated, including transportation, logistics, office and administration support and jobs in the service industry.
Check out this document for clues about whether your job is at risk. The Future Of Employment: How Susceptible Are Jobs To Computerisation? says "about 47 percent of total US employment is at risk."
We examine how susceptible jobs are to computerisation. To assess this, we begin by implementing a novel methodology to estimate
the probability of computerisation for 702 detailed occupations, using a
Gaussian process classifier. Based on these estimates, we examine expected impacts of future computerisation on US labour market outcomes, with the primary objective of analysing the number of jobs at risk and the relationship between an occupation’s probability of computerisation, wages and educational attainment. According to our estimates, about 47 percent of total US employment is at risk. We further provide evidence that wages and educational attainment exhibit a strong negative relation-
ship with an occupation’s probability of computerisation.
My prediction: smarter and more affluent people will more easily accept and productively use automated replacements for many current face-to-face services. The people who will still use, say, bank tellers will be cognitively impaired due to age or born low innate abilities. Anyone who can handle computer interfaces will be able to get goods and services more cheaply.
These researchers expect legal writing and truck driving will get automated. Long haul seems like it'll go robotic before local delivery. The more complex environments of surface streets and more customer interactions make local delivery harder to automate. We could therefore see warehouses located at special highway exits that can handle automated vehicles in controlled environments.
In the present study, we will argue that legal writing and truck driving will soon be automated, while persuading, for instance, will not.
The middle can not hold. Read the bottom of page 12 which ends with:
The result has been an increasingly polarised labour market, with growing employment in high-income cognitive jobs and low-income manual occupations, accompanied by a hollowing-out of middle-income routine jobs.
This is in line with Tyler Cowen's Average Is Over argument.
Look at your job and ask this question: do computers that collect more data enable you to analyze more data and find patterns of greater value? Do you have data analysis skills that get more valuable as more data become available? If yes, you have better employment prospects than most people. But if more data makes your own judgment less valuable and the computer can make better decisions without you then think about how to do a career change that will make your brain more valuable.
Home package delivery: Amazon has a prototype UAV hover craft to do automated home deliveries. Jeff Bezos thinks maybe 4-5 years. Say it is 10 years. Well, 10 years from now lots fewer people will be driving delivery trucks. This will also cut road traffic. Fewer people driving to stores. Brave new world.
Which sorts of service jobs will last the longest? Home repair strikes me as longer lasting than package delivery.
Randall Parker, 2013 December 01 01:57 PM
"The people who will still use, say, bank tellers will be cognitively impaired due to age or born low innate abilities. Anyone who can handle computer interfaces will be able to get goods and services more cheaply."
See, for instance, the lower costs / higher benefits of the internet-only banks (e.g., Ally, ING) versus the traditional bricks and mortar banks (e.g., BofA).
I have come up with the appropriate way to manage this trend. I am an industrial automation/control system engineer.
in the next 20 years machine learning and mobile robotics could even cut into a lot of jobs, including previously safe service jobs...about 47 percent of total US employment is at risk.
A 47% reduction in 20 years is a rate of reduction of 3.1% per year. That's pretty comparable to historic trends for the overall economy, and is slower than the rate of decline of manufacturing employment.
The interesting question is what sectors will grow. I'd like to see medical research grow dramatically. Unfortunately, old sectors like fossil fuels, in their drive to protect themselves from obsolescence, are driving the myth that government and healthcare work can't produce value: healthcare, especially if funded through government, is seen as a cost center, not a profit center, so it's being held back.
I am amused that the discussion does not center on the obvious service job which AI could usefully fill: politicians. I think few would dispute that computers could do a better job than our current legislators, do the work more efficiently ( one report says that Congress spends 50-70% of their time fund raising) and at less cost (no corruption). The ripple effect through the economy would be tremendous: lobbyists would have to find useful work picking up litter alongside the road, the mentally-defect offspring of politicians would be tasked to bring cool water to the parched lobbyists and plush restaurants go out of business since no one will fund $300 bottles of wine on an expense report.
Skynet doesn't have to nuke humanity. It would be much easier to replace the current lot of dolts running the world.
Of course, we would be ruled by programmers but we already are in their power so there should be little change there. Frankly all I see is upside to the change but perhaps some of your readers can see the downside.
That's amusing, but misses the fact that politicians are very effective and efficient at doing their real job: carrying out the wishes of their contributors, while pretending to do what the voters want.
Don't be fooled by the theatre.
It's puzzling - researchers seem to know so little about labor productivity growth. These researchers focus on computerization, but that's just one tool among many for automation and reducing labor inputs.
For instance, music requires great skill and intelligence to compose and perform well, yet player pianos, wax rolls, vinyl platters and cassette tapes almost eliminated the need for local composition and performing. Journalism is a complex cognitive skill, yet improved communications (which use computers, but not AI/machine learning in any direct sense - even Google News aggregates news, it doesn't write it) allowed things like Craigslist and blogging that have decimated their pay and employment.
Washing clothes was automated not by creating a computerized robot with arms, that would take the wash down to the river and beat it against the rocks for you, but by applying electric motors creatively. Furniture making used to be an artistic skill, with both intelligence and fine motor skills, but much of it was replaced by a foot-driven lathe, centuries ago. The list goes on and on.