Artificial intelligence promises to be one of the most important and possibly most disruptive technologies of our time.
It has already permeated every aspect of our daily lives, such as: recommendations when searching, and assisted driving features in cars like Tesla. In the future, AI may help doctors diagnose and fuel the development of fully self-driving cars. From this perspective, AI is a truly epoch-making advancement, considered a "universal technology" like the wheel, the steam engine, electricity and modern factory systems.
Such technologies are a major force in driving entirely new industries, reshaping existing ones, and underpinning the rise of entire new economic systems. Artificial intelligence will have a huge economic impact. It is estimated that AI will add as much as $15.7 trillion to global economic output by 2030. It is no surprise, then, that the U.S. government has placed artificial intelligence high on its list of key technologies affecting innovation, economic competitiveness and national security.
Previously, high-tech industries were concentrated in leading U.S. technology centers and superstar cities (translation: superstar cities refer to the largest and most productive urban areas in the United States, such as New York, the San Francisco Bay Area, Boston, Washington, D.C., and Seattle). Artificial intelligence is likely to reinforce or amplify this pattern of geographic imbalance. These are the key findings of a new study by the Brookings Institution Metropolitan Policy Program (Brookings Metro). The study, led by Mark Muro and Sifan Liu, focuses on the geography of AI. They chart the impact of AI in two key areas: (1) university research, including academic publishing, patents, and federal grants and contracts; and (2) commercialization factors, such as published jobs and workforce skills. Overall, only 10% of U.S. metropolitan areas (36) have a significant AI presence, according to this study. The map below, from this study, identifies five key categories of AI cities and metro areas.
First, the superstars of the superstars. The San Francisco Bay Area, comprised of the San Francisco and San Jose metropolitan areas, is in a category of its own and is the absolute leader in AI in the United States. It is home to leading academic research centers such as Stanford University and the University of California, Berkeley, and a large number of startups, with companies such as Alphabet, Salesforce.com, and Facebook. The San Francisco Bay Area accounts for about a quarter of all AI activity in the United States.
The second is early adopter metros. 13 metro areas with significant AI clusters are referred to as "Early Adopters" by the study. Together with the Bay Area, these 14 metro areas account for as much as two-thirds of the nation's AI assets and capabilities. These include the major cities of the East Coast Acela mega-region - New York, Boston and Washington, D.C.; several California cities, including Los Angeles, San Diego, Santa Barbara and Santa Cruz; and the long-standing hubs of Seattle, Austin, Raleigh and Boulder's long-established central areas. The only surprise is that early adopters also included the smaller communities of Santa Fe, New Mexico, and Lincoln, Nebraska. The former has Los Alamos Laboratory nearby, and the latter has the University of Nebraska.
Third, there are research centers. The Brookings Institution study also identified 21 metropolitan areas with solid research capacity but little commercialization, including: Pittsburgh, home to Carnegie Mellon University and the University of Pittsburgh, and smaller college towns such as Ann Arbor, Madison and Durham-Chapel Hill. Durham-Chapel Hill is adjacent to Raleigh and is part of the Research Triangle Park (Research Triangle).
Potential Artificial Intelligence Centers. This group consists of 87 metropolitan areas. The study refers to them as "potential adoption centers" with more moderate AI activity. In total, these cities generate a quarter of all AI patents and companies in the U.S. and account for a third of all AI jobs and workers. But on a per capita basis, these cities have less than half the AI capacity of the 13 early-adopter cities. They include fast-growing "Sunbelt metros" like Atlanta, Houston and Nashville, and "Frostbelt metros" with large industry clusters like Chicago and Detroit. Frostbelt metro areas, college towns like State College, home of Penn State University, and smaller tech hubs like Provo, Utah.
The trajectory of inequality
Many regions in the U.S. are hoping that their attractiveness to companies and talent in emerging industries like artificial intelligence will increase as the epidemic causes a shift to remote work and tech companies move more to the cloud rather than physical locations. A growing number of states and cities are developing strategies and initiatives targeting AI.
The data provided by Murrow points to a potentially promising sign: a significant increase in AI-related jobs in several metro areas in 2020, and a slight decline in the Bay Area. The likely scenario is that the geography of AI technologies, jobs or startups may not change significantly in data for several years.
So far, however, the rise of AI has followed the spike pattern we have seen in technologies such as semiconductors, software, biotechnology, the Internet, social media and cloud computing, where new technologies and industries grow around a few central areas of dominant technologies.
This is evident from the fact that the Brookings Institution study found that the vast majority of U.S. metropolitan areas (260), have virtually no significant AI capabilities. This fact is quite surprising. In fact, Pittsburgh's experience demonstrates the difficulty of creating new technology centers. Carnegie Mellon University (where I taught for nearly 20 years) has long been one of the top three centers for AI research in the United States, alongside Stanford and MIT. The study puts Pittsburgh on the list of research centers with limited entrepreneurship or commercialization.
Federal intervention may be needed to counter and reshape the powerful trends underway
Artificial intelligence is different from previous high-tech fields in that its rise is not just happening in the United States, but globally. It is growing in a series of global tech hubs such as London, Berlin, Tel Aviv, Shanghai, Beijing, Bangalore, Montreal and Toronto. In fact, the Brookings Institution study mentions Toronto's Vector Institute, a consortium of universities and global companies, as an example that regions can look to in building their AI clusters and ecosystems. My research with Ian Hathaway charts the rise of global tech hubs against the backdrop that the U.S. share of global venture capital going to startups has fallen from more than 90 or 95 percent in the 1990s to less than half today. This rise of global tech hubs is likely to increase in the future only if the new crown era of travel, immigration restrictions continue, and many of the world's cities become more attractive as gathering places for both homegrown and global talent.
In addition to these geographic challenges, AI is expected to have a dramatic impact on employment, potentially eliminating not only a large number of low-skilled jobs in manufacturing and services (e.g., transportation, logistics, and retail), but also in professional and knowledge work areas (e.g., medicine, law, and engineering). Given this threat to jobs and existing industries, the Brookings Institution study encourages all regions to carefully assess the threats and opportunities posed by AI to their economies and workforces. This, in turn, is very important.
Federal intervention may be needed to counter and reshape the powerful trends underway. The Innovation and Competition Act of 2021, proposed by the Biden administration and being considered by Congress, would target $250 billion in investments in key technologies such as artificial intelligence to drive the development of new "regional technology hubs" (RTCs) across the United States.
If AI is allowed to grow freely, it will reinforce and exacerbate the "winner-take-all" attributes of the economic and geographic landscape.
