This article is an interesting look at a recent AI conference. It basically sums up the interesting pints of the conference. The whole thing is good, but this is what caught my eye:
“I’d suggest… whenever AI comes in, even potentially to replace labour, it’s genuinely because it’s an efficiency gain — so creating more. But then perhaps the way to think about it is how this efficiency gain is distributed. So if it’s concentrated in the hands of the owners perhaps that tends to be not of good value to society. But if the benefits accrue to society at large that’s potentially better,” said Dalyac.
“For example something that we’re working on is automating a task in the visual assessment of insurance claims. And the benefit of that would be to lower insurance premiums for car insurance… so this would be a case where the people who are usually employed to do this would find themselves out of work, so that might involve maybe 400 people in this country. But as a result you have 50 million people that benefit.”
That is the standard answer whenever the prospect of work being lost to artificial intelligence and automation in general. This is a limited answer, almost glib. I know it is not meant to be, but it is. It glosses over several points and assumes some facts not in evidence.
First, the numbers seem to be completely out of whack. It doesn't seem likely that only 400 people in this country visually inspect insurance forms meaning that the ratio is not quite as rosy as it seems. Even if it is only 400 people, if there exists technology that can parse a complicated insurance form, it can do other, similar work in other industries, driving out more workers.
Second, there tradeoff is only mentioned in terms of workers losing their jobs and it appears to be a simple math equation. If the number of workers affected is lower than the number of consumers that benefit, then it is good. But of course, every time a person loses their job, the economic impact radiates out. What good is a lower insurance premium to the clerk who lost their job because the disruption in industry meant a loss in business for his store.
Third, it is not guaranteed that the benefits of automation will accrue to consumers. Outside of textbooks, prices do not go up and down in perfect unions with costs and savings. It is surprisingly hard to find information about the overall benefits of automation. We do know, however, that the economic benefits as a whole have gone to the top of the economic pie, sometimes the very top. We also know that in the recent recession, and entire class of jobs seems to have been destroyed:
Nevertheless, the recession of 2007–2009 may have sped up the destruction of many relatively well-paid jobs requiring repetitive tasks that can be automated. These so-called routine jobs “fell off a cliff in the recession,” says Henry Siu, an economist at the University of British Columbia, “and there’s been no large rebound.” This type of work, which includes white-collar jobs in sales and administration as well as blue-collar jobs in assembly work and machine operation, makes up about 50 percent of employment in the United States. Siu’s research also shows that the disappearance of these jobs has most harshly affected people in their 20s, many of whom seem to have simply stopped looking for work.
This all suggests that society does not benefit by automation at anywhere near the equal levels to make it a net positive, or even net neutral, phenomenon.
I am not a luddite. I program for a living, for crying out loud, and I truly believe that in a perfect world that automation and artificial intelligence is a huge benefit for humanity. But that assumes that humanity has a social structure that allows for people to do less and less work and still live full and enriched lives. That has not happened with the advances so far. More and more of the wealth has gone to the wealthiest among us. With a political culture that cheers people dying if they cannot get insurance, where half of the major political parties are dominated by the notion of "moochers vs makers", where the majority of economic benefits go to a smaller slice of people higher up on the economic ladder, there is precious little reason to believe that automation will help advance such a society.
Automation and artificial intelligence will continue to advance, and they will almost certainly continue to work into more and more kids of jobs, jobs we used to be told couldn't be done by machines. We have already seen how that can destroy entire classes of jobs, making it very hard for those people to recover. We know that this kind of disruption can be enormously difficult to overcome:
In his new book, The Great Divide, the Columbia University economist Joseph Stiglitz suggests that the Great Depression, too, can be traced to technological change: he says its underlying cause was not, as is typically argued, disastrous government financial policies and a broken banking system but the shift from an agricultural economy to a manufacturing one. Stiglitz describes how the advent of mechanization and improved farming practices quickly transformed the United States from a country that needed many farmers to one that needed relatively few. It took the manufacturing boom fueled by World War II to finally help workers through the transition. Today, writes Stiglitz, we’re caught in another painful transition, from a manufacturing economy to a service-based one.
Half answers and platitudes about
“Short-term and medium-term, [AI] will displace work but not necessarily jobs,” he says. If AI tools do some of the scut work involved in analyzing data, he says, people can be “free to work at the top of their game.”
Won't cut it. If you aren't working actively toward concrete solutions, then you are defaulting to the dystopian future, with a mass of under or unemployed people left behind and a very small upper class controlling the vast majority of the gains from Ai and automation.