Bloomberg: Artificial intelligence can create jobs, too. How else do we manage the robots?
Artificial intelligence can create jobs, too. How else do we manage the robots?
You’ve probably heard about that 2013 study by two University of Oxford researchers that found 47 percent of U.S. jobs, most of them in the service sector, to be at high risk of “computerisation” over the next decade or two.
Other researchers have since argued that this overstates the risk because, while certain tasks are likely to be automated, most jobs involve multiple tasks. A study out last month from the Organization for Economic Cooperation and Development, for example, found that just 14 percent of the jobs in 32 affluent OECD-member countries studied were “highly automatable,” although another 32 percent were likely to see “significant change in the way these jobs are carried out as a result of automation.” For the U.S., which is already further along in automation than OECD members such as Slovakia and Greece, those percentages were 10 and 28.
I read Daugherty and Wilson’s new book, “Human + Machine: Reimagining Work in the Age of AI,” on the way to a Bloomberg Ideas conference in San Francisco last month, where I and several of my Bloomberg View and Gadfly colleagues were going to discuss the economics of artificial intelligence. It is to some extent the kind of book one would expect from the chief technology and innovation officer (Daugherty) of a major information-technology and management consulting firm and the managing director of information technology and business research (Wilson) at the firm’s research arm. There are bullet-pointed lists. And flow charts. And sentences like:
In spite of all this, though, it is also an engaging and interesting book. Especially interesting to me was the chapter in which Daugherty and Wilson explore 13 “entirely novel positions, requiring skills and training never needed before,” that AI will create. Without further ado, here they are, grouped into three categories — trainers, explainers and sustainers:
- Trainers: empathy trainer, personality trainer, worldview and localization trainer, interaction modeler, data hygienist.
- Explainers: algorithm forensics analyst, transparency analyst, explainability strategist.
- Sustainers: context designer, AI safety engineer, ethics compliance manager, automation ethicist, machine relations manager.
What will an empathy trainer do? “Teach AI systems to display compassion.” Data hygienists will ensure that the data used to train AI systems is “free from any slanted perspective.” Explainability strategists will be “responsible for making important judgment calls about which AI technologies might best be deployed for specific applications.” Machine relations managers “will function like HR managers, except that they will oversee AI systems, not human workers.” And so on.
No, it wasn’t entirely clear to me even after reading the entire chapter what exactly these jobs would entail or what kind of training would be needed to do them. But I was struck by the common theme of humans guiding, correcting and managing the robots. This is something that some of the companies pioneering the use of artificial intelligence in recent years — Alphabet Inc., Amazon.com Inc., Facebook Inc., Uber Technologies Inc., etc. — have not been great at. “Part of the issue is that the tech companies haven’t done this, which has gotten us to where we are now,” Daugherty told me last week. “The concerns we’re seeing are a teachable moment for organizations. These are the things everything is going to face.”
Another thing that struck me was that some of the jobs felt kind of transitional. At some point, one imagines, the AI systems will get so good that they presumably won’t need training in empathy or personality. Daugherty thinks that moment is pretty far off, though. “You’ll need roles of these types for a very long time to come,” he said.
His own industry provides a nice example of that. Computers were supposed to take over a lot of clerical and even management roles in business, and to some extent they have. But this has created tons of jobs for the people who design and manage the computer systems and advise organizations on how to use them. Daugherty, for example, said that about half of Accenture’s 435,000 employees are doing work that didn’t exist as recently as five years ago.
Here is what has happened since 1990 to U.S. employment in two services sectors that more or less cover what Accenture and its rivals do (computer systems design and management and technical consulting) and two that you might expect to have lost some jobs to automation (accounting and law).
Accounting and legal services have indeed added a lot fewer jobs since 1990 than the other two, and legal services has shed jobs over the past decade. But overall there are 5.6 million people working in these four sectors now, compared with 2.3 million at the beginning of 1990. This representation isn’t perfect; in some cases, the IT and management consulting firms have simply taken over tasks that were once performed in-house by companies in other industries. 1 Also, past performance is no guarantee of future results. Still, it is at least evidence that automation can create jobs, too.
- Annual data on jobs, as opposed to industries, is available from the Bureau of Labor Statistics’ Occupational Employment Statistics, but the BLS cautions against comparing numbers over time because of “changes in the occupational, industrial, and geographical classification systems, changes in the way data are collected, changes in the survey reference period, and changes in mean wage estimation methodology, as well as permanent features of the methodology.” In any case, in 2017 there were 4.8 million Americans working as business operations specialists of various sorts, 4.2 million in computer and mathematical occupations, 2.7 million financial specialists (including accountants and auditors) and 1.1 million in legal occupations.