AI & The Changing Face of Work
Work consumes something like two-thirds of our waking hours, so the shape of our workplace, content of the work, and skills required for work matter tremendously to our future. While these factors usually change slowly, COVID has forced a dramatic shift in a short time. Meanwhile, new, and sometimes scary forms of automation, in particular “artificial intelligence” (AI), promise to change the nature of work even more. The future of work is thus both a timely topic and a long-term challenge. Harvard Business School professor Joseph Fuller, whom I spoke with in September 2022, looks at the future of work from the perspective of a veteran scholar who has also consulted to many large corporations. In addition to his academic career, Fuller founded the Monitor Group (now Monitor Deloitte) with fellow HBS professor Michael Porter, author of the bestselling book Competitive Strategy. Fuller focused on (1) the changing skills mix of the U.S. workforce, (2) the inefficient way we screen job applicants, (3) credentialism and the need or lack of need for college, (4) retooling the education and training system for the work requirements of the future, and (5) the changing role of women in the workforce. That’s a big bite to chew on, but a small subset of the overall topic. Among the topics that were undiscussed or lightly discussed are the competitive positions of different countries and the recent sudden disruption of the global supply chain. The future of work and the future of business and society are not quite coextensive, but they’re close.
The Changing Skills Mix of the U.S. Workforce
Let’s briefly recap what Fuller said about each of the five main topics. He began by reporting results of a survey of HBS alumni: “the workforce in the United States had been a significant source of competitive advantage historically but that that advantage was waning rapidly.” This is not a good thing. Because we’re one of the world’s most advanced countries, it’s natural that we face catch-up competition; however, the United States cannot afford to become a laggard in human capital. Other than a nice piece of real estate, human capital is all we’ve got. And the much-discussed mismatch between the skills needed by employers and the skills workers have (or don’t have) is visibly getting worse.For example, at the high end, we teach calculus to high school seniors. Only those who later become certain types of engineers will ever use calculus; everyone else regards it as torture. Meanwhile, we fail to teach probability and statistics, which are the backbone of AI and, for that matter, of all decision-making under uncertainty.The job mix is changing too Along with the changing mix of skills, the mix of job types has changed. (These two changes are not in harmony, hence the friction we’re feeling – lots of people working at jobs for which they’re overqualified, and many other jobs being done poorly because the workers are underqualified.) Using the Fed’s classification scheme for job quality – nonroutine cognitive (the highest paid), routine cognitive, nonroutine manual, and routine manual – Exhibit 1 shows the number of jobs in each category from 1983 to 2022. There are now about 68 million nonroutine cognitive (“thinking”) jobs, about 48% of all jobs. In addition, there are 26 million nonroutine manual (“craft”) jobs that also involve thinking.[1] Meanwhile, routine jobs have stagnated, and as a percentage of the working population they have declined.
Exhibit 1
Number of jobs by quality category, 1983-2022, Federal Reserve classification scheme
Source: FRED database, Federal Reserve Bank of Saint Louis
Applicant screening folly
While it’s easy to blame the educational system and the workers themselves for the mismatch, Fuller also blames the employers: “a lot of the problems companies were observing, and frankly complaining about, were often substantially of their own making.” He then launched into a diatribe on the ridiculous ways that employers narrow a large applicant pool to one small enough to be practical. Before COVID, companies got an average of 250 applicants for a middle-skill job – say, one paying $60,000 per year – and would need to narrow that candidate population to 3 or 4 interviewees. (The current employee shortage has made the problem more tractable; but don’t get used to it.) To accomplish this daunting task, they use a technology called an applicant tracking system, which is based on AI. Fuller commented, “I don't really think it's all that intelligent. It uses very blunt proxies to evaluate the attractiveness of workers.”For example, some (human) “expert” told the AI system, presumably based on a study of what makes a job candidate desirable, that continuity of employment is very important. That criterion screens out everybody who’s interesting – people who started a business that didn’t work out, had kids, cared for an aging parent, studied abroad, traveled with a sports team, or did a tour of military duty. A company consisting entirely of super-straight arrows (who also never had any bad luck) would not just be a terrible place to work; it would be terribly uncompetitive.
Is AI going to displace knowledge workers?
Nonetheless, Fuller says that “AI is making unbelievable progress in non-routine cognitive jobs” – the ones we’ve assumed could not be automated because they require creativity and judgment. Fuller presents some evidence of that progress, but I’m skeptical. I’m fine with a robot filling out tax forms or diagnosing what’s wrong with my car, but I want a human being (even if he is in India) interpreting my radiology results.
Should everyone go to college?
On to credentialism. In 1960, 7% of adults had a college degree. Today, 28% do. Based on the behavior of the youngest cohort of adults, that number will approach 40% as that cohort becomes the dominant age group in the workforce.Is this a good thing? My bias is that more education is better than less, and with the percentage of cognitive jobs already huge and still rising, we need more people trained formally in cognition. But not everybody is capable of college work, and not every job should be filled by someone who is.Meanwhile, we have become a nation where you are what you can hang on your wall. Credentials are everything. Raw talent, creativity, personality, ambition, and “grit” count for little if you’re being evaluated by an AI program that first screens out non-college graduates and then looks at your other qualifications.There also needs to be a place in our economy for people without all those wonderful attributes. Average people need jobs too, and can produce much that is valuable.Fuller sees some potential in imitating the practices of other countries: “We need to do as Germany has done. For years they've celebrated non-college occupations: plumber, electrician, construction worker, automotive engineer.” I applaud this suggestion but see one problem with it: what will they do after they turn 55? They can’t all supervise, and I don’t want an 80-year-old roofer working on my roof.Fuller also notes a downside to the German approach: that the country is falling behind in developing digital talent. Cars are becoming computers with wheels. Automotive engineers of the future will need to have very sophisticated knowledge of the digital world.
The role of women
Finally, most kinds of work have gotten physically easier, more intellectual, more flexible in terms of time requirements, and more rewarding of diligence and cooperation. Every single one of these trends favors women. It is thus unsurprising that women are becoming a larger share of the workforce, are holding more high-paying jobs, and (interestingly) are holding fewer low-paying ones.[2] What’s behind this? “Women materially and widely outperform men on various measures of social skills,” said Fuller. How about intellectual skills? “Of the population of students in higher education, 58% are women.” It’s true that men still dominate in engineering and the hard sciences, but even that edge is diminishing.Fuller concludes that “the future in many high value-added jobs is increasingly female. Expectations about the content of work will have to change considerably to accommodate the reality that more and more of the most qualified people are women. Having conditions of employment that accommodate their broader caregiving and other obligations will be absolutely essential.”This outcome, which I welcome (although negative effects on men should be taken seriously), is the predictable result of technology improving the quality of work. We should have seen it coming sooner.
Listen to our full conversation here: Joseph B. Fuller: The Future of Work
[1] “Thinking” and “craft” are my terms, not the Fed’s.
[2] https://libertystreeteconomics.newyorkfed.org/2020/03/women-have-been-hit-hard-by-the-loss-of-routine-jobs-too/