Google Chief Economist Weighs Bots vs Tots
By Robert Bernstein
"Google Chief Economist" is a title that gets attention when Hal Varian speaks. He was a keynote speaker at a recent UCSB conference "Mind and Machine Intelligence".
The original title of his talk was "Automation vs Procreation". But he shortened this to "Bots and Tots".
Varian explained that he is a theorist. The basic question at hand is about supply and demand for labor. Fewer children in recent years in the US (procreation/tots) means less supply of labor. Which raises wages.
At the same time, more robots (automation/bots) means less demand for labor. Which lowers wages.
This slide shows this:
The question he is trying to answer: Which of these two effects is bigger? Is the decline in having children going to lead to a shortage of workers? Or is the increase in automation going to lead to a shortage of jobs?
He started with some historical examples. Going back to this 1812 Wanted poster. Workers had destroyed a cloth making machine to try to keep their jobs from being "stolen".
He showed a series of slides from 1935, 1960, 1980 and 2016 with the same theme of automation "stealing jobs".
After World War II there were two shocks to the labor market. First was the Baby Boom 1946-1964. Then there was a baby bust. Followed by a Baby Boom Echo.
The second shock was the entry of women on a large scale into the labor market.
Now we have an era of Baby Boomers retiring. And the number of women in the workforce has plateaued.
The most interesting point of his talk came next: He noted the difference between Jobs and Tasks.
There were 270 detailed occupations listed in the 1950 US Census. Only one has been eliminated due to automation. That one job? Elevator Operator.
Even that job had other tasks besides operating the elevator:
- Safety and Security Monitor
- Answering Questions
These tasks and others were added to the responsibility of other jobs like receptionist and security guard. A Job is a collection of Tasks. Automation doesn't replace entire jobs in general. Most jobs are more complex than we think. Whether the job is primarily manual or cognitive.
It took two slides for him to display all of the tasks that a groundskeeper is responsible for.
You could automate any one task. You just need a billion dollars and a squad of engineers! Eliminating all of them is not practical.
Robots work best for repetitive tasks. Such as assembly lines.
Henry Ford got the idea for the assembly line when he visited meat packing plants in Chicago. They had already perfected the idea of an animal carcass being moved along on a "dis-assembly" line of workers carving off their selected pieces.
Half of all industrial robots are now in automobile plants because of the repetitive nature of the work.
A heterogeneous environment like a groundskeeper deals with is not easily automated.
We often talk about a need to create "good jobs". But Varian asked us to consider: People do gardening as a hobby. Do you know anyone who does assembly line work as a hobby?
Hotel Housekeeper is a similar challenge as Groundskeeper when it comes to automation.
Varian grew up on an orchard in Wooster, Ohio. Orchards today are not like those. Apple trees now are the height of a person. And the apples are only growing on one side of the tree! Which makes for something like an assembly line for apples.
But this is much harder for delicate fruits like berries or peaches.
Can we replace humans with humanoid robots? Like C-3PO in Star Wars?
Our household machines don't work like humans. Think of dishwashers, clothes dryers, vacuum cleaners or sewing machines. "Cars don't walk."
The amount of non-routine work is growing for both manual and cognitive categories, Varian claimed.
Henry Ford said, "In mass production there are no fitters." Before his production lines, workers had to file parts to get them to fit.
But Elon Musk discovered in 2018 that excessive automation at Tesla was a mistake. Musk declared that "Humans are underrated."
What can be automated? Estimates vary widely and Varian showed this simple chart of some of those estimates:
The MIT magazine Technology Review offered a table of such wide-ranging estimates in January 2018. That article ends with this statement:
"In short, although these predictions are made by dozens of global experts in economics and technology, no one seems to be on the same page. There is really only one meaningful conclusion: we have no idea how many jobs will actually be lost to the march of technological progress."
Here is that article in full: https://www.technologyreview.c
The top ten largest occupations in the US are all service jobs: Retail salesperson, cashier, food preparation, office clerk, registered nurse, customer service representative, waiter/waitress, laborer, administrative assistant and janitor.
Many tasks could be automated. McDonalds now has an automated ordering system in some of its restaurants.
Suppose we did increase productivity by 25%? We could work just four days a week. Or retire younger. Or consume more.
It is just a social convention to work an 8 hour day and 5 day week. Mexico has a 45 hour week. The Netherlands has a 29 hour week on average.
What do people want? More jobs and less work. Technology can deliver that. A Universal Basic Income some love, some hate. But most people love the idea of a three day weekend.
How do we get to more jobs and less work? It will take education and training. Higher education is associated with higher pay and lower unemployment.
Varian asked if this is an example of the "Fallacy of Composition". Is it good for everyone to be more educated? After all, if an individual stands up at a rock concert, she may get a better view. But if everyone stands up, the effect may not be so good.
Varian answered this by saying that even jobs like groundskeeper and housekeeper will involve tasks that will be automated. And that means it is important to learn the skills to use that automation.
How best to learn those skills? One way is to learn on the job. But Varian noted that there are a billion views each day on YouTube of "how-to" videos. Everything from how to change a tire to how to bake a cake.
If school kids miss that math class on quadratic equations, they can view it online. This does not mean that teaching jobs go away. It just provides a supplement to the work that teachers provide.
Varian went on to talk about "Cognitive Assistance". It used to be that a cashier had to know how to make change. A taxi driver had to know all the streets. Now there is "cognitive assistance" for these tasks and many other common tasks. This assistance does not eliminate jobs. On the contrary, it allows people to do jobs that they may not have qualified for in the past.
Then Varian got into the second half of the talk: Tots/Demography.
Productivity is the measure of output per hour worked per employee. For society as a whole, "participation" is a factor in determining total output: What percentage of the population is in the workforce.
Right now we are at full employment. And productivity is not growing very fast in recent years, he claims.
And "participation" is declining as the population ages.
The 2020s will have the lowest growth of the labor force in US history. It will stay that low for the next 30 years. In fact, without immigration, the labor force would be shrinking in the US.
He talked about the "Dependency Ratio". This represents the ratio of workers to those who are dependents or are otherwise not working. Japan is facing a serious crisis. By 2050 about 80% of the
Japanese population will be over 65.
In contrast, the US is doing better than many industrialized countries in this respect. By 2050 about 40% of the population will be over 65.
But the labor shortages in the US will not be evenly distributed geographically. He showed this graphic.
China is an interesting case. They ended their one child policy. So the birth rate is rising. But it still takes 20 years to produce a 20 year old! Which means many in China are too young to work for now.
The US birth rate is at an all-time low. Why is the US birth rate so low? Varian points to economics. Child care being too expensive is the top reason people say they don't want to have children.
Varian showed a chart of industrial robots by country. China is huge. Far and away bigger than any other country.
So, Bots vs Tots: Which is the bigger effect? Varian argues that Tots/Demographics is the bigger effect for the foreseeable future. We will face a labor shortage because population growth is low and the population is aging.
As retirees age, they become even more costly. There are 5 million with Alzheimers in the US today. That number is expected to balloon to 14 million in 2050.
Varian then answered some questions.
He noted that different countries handle these challenges differently. The Dutch system allows parents to work less and thereby take turns caring for their children.
A Dutch participant in the conference noted that, "We did Bernie 60 years ago!"
Varian noted that a declining population has many advantages, even if it creates medium term challenges. A lower population lowers environmental impacts, most importantly.
One participant was concerned that we might lose cognitive skills with automation. Varian did not seem so concerned. He noted that he had to learn how to calculate square roots in school. No one learns that skill anymore and he thinks that is not so bad.
He said it is more important to use the technology effectively. People may do square roots or other calculating challenges as a hobby. The same way people care for horses, even though horses are no longer needed for farm work or transportation.
He thinks that our education system is overdue for realignment with current needs. He thinks learning statistics is more important than learning trigonometry. And he thinks that everyone should learn some ability to write computer code (programming).
I will note that when I was in 8th grade in the 1970s our entire junior high class learned to write simple programs in BASIC. I have been astonished to learn that more recent generations have not had such a fundamental education in computer programming.
"It’s Difficult to Make Predictions, Especially About the Future". The origin of this quote is not known for sure. But it is important to keep this in mind. If "Artificial General Intelligence" (AGI) becomes a reality, that would change everything. It would allow for robot groundskeepers and housekeepers. And once one robot has mastered the job, they could be duplicated. It does not take 20 years to make a robot copy that has the experience of a robot with 20 years experience.
Varian left us with some valuable points that show that automation is less of a threat to jobs than might be thought. And he showed that, at least in the medium term, we may be facing labor shortages rather than unemployment. It seems that in the battle between Bots and Tots, the winner is Tots.