Crossing the Great AI Divide

Technological breakthroughs often lead to structural changes in the labor market. The power loom, the internal combustion engine, the computer—each remade or replaced some once-critical occupations (for instance, who has met a lamplighter?). Yet as painful as these changes can be, new occupations rise from the ashes to replace the old. To some, the specter of AI suggests that the end of human work is on the horizon, and with it, the end of postsecondary education as job preparation. I think it’s more likely that both human work and postsecondary education will survive, albeit transformed.

My views on how AI will affect human work stem from how occupations are structured. Occupations don’t reflect just one activity but rather consist of a broad portfolio of tasks. While the name of an occupation, like accountant, may reflect the primary task it involves, the idea that this task is the whole of the occupation is antiquated, tracing back to Ford’s assembly line and Taylor’s principles of scientific management. Most occupations require a multifaceted collection of skills rather than repeated application of a single skill. For instance, plumbers need carpentry skills, electrical skills, accounting and estimating skills, and personal skills to deal with clients—something I know firsthand from my previous life as a general contractor. 

Research demonstrates just how important it is to understand occupations as a comprehensive mix of tasks rather than as a single task. For example, in “Revisiting the Risks of Automation,” Arntz and colleagues found that research that considered only an occupation’s main task greatly exaggerated the risk of automation to US employment. When estimating the risk that automation poses to jobs by associating occupations with a single task, they found that 38 percent of US jobs were at risk of elimination due to automation. In contrast, when estimating the risk by accounting for the full scope of tasks within occupations, they found that only 9 percent of US jobs were at risk of elimination.

The research on automation holds lessons for what I envision will be one impact of AI. Like automation, AI will likely change how we do business but will not fundamentally change the problems that most businesses have set themselves to resolve, nor the activities they undertake. Regardless of how pervasive AI becomes, we will still need humans to oversee the programming, accounting, and business analysis tasks that AI can help streamline. At least some part of the current AI boom is almost certainly “smoke and mirrors.” But once AI becomes more trustworthy, I suspect that human jobs will increasingly involve using AI to accomplish work activities—when and as appropriate. 

If my suspicion is correct, AI literacy—the judgment to determine when and how to use AI tools effectively—may be the critical skill graduates will need to succeed in the emerging labor market. Fortunately, it’s a skill that US colleges and universities already have the foundation to provide. Since long before the emergence of AI, US postsecondary education has balanced general and specific education, teaching students both how to master specific topic areas or fields of practice and how to adapt their skills to unfamiliar domains. Historically, this adaptability has enabled the US workforce to be resilient in the face of global change while giving the bachelor’s degree staying power in a mutable labor market. It has also increased the importance of college credentials as technology-driven changes in business practices make the workplace more complex. Yesterday’s auto mechanic is today’s automotive technician: Workers who once depended on mechanical skills must be computer literate to repair cars that are increasingly computerized.

Of course, it’s hard to really know how AI will affect the future economy. Current research is all over the map, making it very difficult to model potential shocks to the system. We have some robust evidence that entry-level employment in AI-exposed occupations has fallen precipitously. But what we don’t know is how these occupations, and the educational qualifications they require, will morph as the labor market and AI use evolve. And yes, the labor market is increasingly difficult for recent college graduates, but the economy is slowing for everyone—not just those with bachelor’s degrees. Overall, the evidence suggests that AI isn’t causing a unique “jobs apocalypse.” Instead, it is changing occupations at the task level, potentially eliminating certain jobs in a way that mirrors the transformation of the modern workforce following the introduction of other new technologies (namely, computers and the internet). 

While we cannot predict the future, what I do know is that our education system builds self-reliant individuals who should be able to adapt to new technologies quickly. In fact, graduates who know how to work with AI will likely have an edge in the workforce. Even more important, those who know how to work with AI responsibly and ethically can help prevent people’s worst fears about AI from coming to pass.

However, much still needs to happen for AI literacy to effectively undergird the bridge between the workforce of today and the workforce of the future. Our colleges need to develop best practices and good policies for AI use, and technology companies need to ensure that AI delivers on its promises. But whatever happens next, I remain optimistic about the role US colleges and universities can play in helping people prepare for it.