A key takeaway of a report on AI and automation released in 2019 by the Brookings Institution was “the vulnerable will be the most vulnerable” and that higher-wage, better-educated workers will largely make out all right as automation spreads. But what about artificial intelligence (AI), the increasingly powerful form of digital automation using machines that can learn, reason and act for themselves?
A new report from Brookings uses a method developed by Stanford University Ph.D. student Michael Webb to identify the kinds of tasks and occupations likely to be affected by particular AI capabilities to do further analysis of which types of jobs and which areas of the U.S. could be most exposed to AI. “Where the robotics and software that dominate the automation field seem mostly to involve ‘routine’ or ‘rule-based’ tasks (and thus lower- or middle-pay roles),” write Mark Muro, Jacob Whiton, and Rob Maxim of Brookings, “AI’s distinctive capacities suggest that higher-wage occupations will be some of the most exposed.”
|San Jose-Sunnyvale-Santa Clara, CA
|Grand Rapids-Wyoming, MI
|Louisville/Jefferson County, KY-IN
|Salt Lake City, UT
|Greensboro-High Point, NC
|El Paso, TX
|Cape Coral-Fort Myers, FL
|Las Vegas-Henderson-Paradise, NV
|Deltona-Daytona Beach-Ormond Beach, FL
Where will AI potentially inflict the most damage or, conversely, have the least impact? Here’s their analysis of AI’s geographic footprint:
An initial look at the state-level exposure map suggests, right off, that the nation’s eastern heartland — sweeping from Wisconsin and Michigan though Indiana, Kentucky, and into Alabama and Georgia — will be heavily involved with AI given its association with manufacturing, which is increasingly linked with machine learning and related applications. At the same time, significant additional exposure can be discerned along the high-tech and managerial Boston-Washington, D.C., corridor as well as in Washington state and California.
Much of this map is familiar to earlier mappings of automation’s impact, with two exceptions. Nevada has flipped from being one of the most exposed states to one of the least, since AI is much less likely to disrupt the accommodation and food services sectors. Conversely, Washington state has moved in the other direction and is highly exposed to AI, which surely has to do with its specializations in both advanced manufacturing and technology in and around Seattle.
Looking now at community types, more reversals surface. Contrary to the automation maps, for example, the present AI analysis reveals that smaller, more rural communities are significantly less exposed to technological disruption than larger, denser urban ones. This likely reflects the basic urban geography of the information, technology, and professional-managerial economy, with its orientation toward analytics, prediction, and strategy — all susceptible to AI solutions.
Similarly, bigger, tech-focused metro areas and manufacturing hubs dominate the list of highly exposed larger places, and the full list of exposed metro areas displays a number of smaller manufacturing or agricultural places as well.
Among the most AI-exposed large metro areas are San Jose, California; Seattle; Salt Lake City and Ogden, Utah — all high-tech centers — along with agriculture and logistics hub Bakersfield, California, and manufacturing centers Greenville, South Carolina; Detroit; and Louisville, Kentucky. Filling out the high-exposure end of the full list are manufacturing places (Elkhart-Goshen, Indiana; Dalton, Georgia; and Columbus, Indiana), agricultural centers such as Madera and Salinas, California, and high-tech concentrations including Boulder, Colorado, and Huntsville, Alabama.
Places that appear most disconnected from AI are heavily concentrated in the Sun Belt. They range from bigger, service-oriented metro areas such as El Paso, Texas; Las Vegas, and Daytona Beach, Florida, to smaller, “leisure” communities including Hilton Head and Myrtle Beach, South Carolina, and Ocean City, New Jersey. These metro areas lie far from manufacturing and technical-managerial regions, and focus on providing low-tech, AI-free interpersonal services to the leisure class.
Exposure doesn’t necessarily mean substitution. AI may well complement human work in many of the highlighted occupations and regions. But it will likely be disruptive and require change and adjustment.
For the full analysis, visit brookings.edu.