Measuring the Future: AI and San Diego's Economy

San Diego Regional Economic Development Corporation, in partnership with Booz Allen Hamilton and the San Diego artificial intelligence (AI) community at-large, launched a study to define AI and Machine Learning (ML), identify key industries and clusters in San Diego County where AI and ML have been implemented, and ultimately quantify the impacts of these technologies on the San Diego regional economy.

San Diego's AI economy is in a strong position. Contrary to popular belief, AI and ML technologies are creating jobs, not destroying them, and this is true for AI developers and adopters alike. More than three in five AI developers plan to add additional workers in the next year as demand for these technologies grows. These jobs tend to be lucrative and will have large ripple effects on the broader economy.

Also encouraging, firms operating in industries that typically serve as adopters of AI and ML technologies have reported an increase in productivity as they've undergone implementation. The growth implications for the region are positive, since each $1 billion in additional Gross Regional Product (GRP) from productivity gains in these industries translates to just over $750 million in additional growth in other industries. To this point, the vast majority of respondents in a business survey conducted for this study reported that AI and ML technologies are supercharging business activity across a number of metrics, including improved productivity, increased revenues, reduced costs, and new product development.

The data support the idea that local firms engaged with AI are using these technologies to complement the workforce rather than replace it. Research suggests that business outcomes are much more positive for companies that use AI as a tool to enhance their current workforce, putting San Diego on a firm footing when it comes to AI proliferation.

Finally, San Diego-based AI technology could have widespread appeal in other U.S. markets where companies are struggling to source qualified workers for open roles. San Diego AI developers are producing technologies in a number of specialty areas with commercial viability here at home and across the country (e.g., New York City, Detroit, etc.). Importantly, these technologies could be implemented without disrupting local job markets, since they would be used to substitute unfilled positions, not jobs that are already filled.

AI and ML technologies are swiftly becoming more commonplace as computing power and speed increase. Self-driving cars, algorithmic trading, customer experience bots, and AI assistants, like Siri and Alexa, have become commonplace tools used by people at home and at work.

 Estimates  place the boost global productivity growth from AI and ML somewhere between 0.8 to 1.4 percentage points, annually, depending on how quickly businesses adopt the technologies in the coming decades. This translates to 45 to 90 trillion dollars—roughly the equivalent of two to four U.S. economies—in additional global output over the next 30 years.

Given its complexity, a common definition for AI has been elusive. For the purposes of this study, AI is defined as, “computer-based automation that seeks to simulate or enhance human decision-making.”

The advent of AI and ML technologies in our daily lives promises to be a transformative force for businesses worldwide. This report seeks to establish a baseline understanding of AI’s current reach in San Diego, in terms of development, investment, and adoption. In addition, the report outlines the potential future impacts of AI through the lens of, both, productivity growth and the labor force.

This report is the first in a series that will explore the proliferation of AI in San Diego from a number of angles, including deep-dives into the industries where AI and ML have made the most significant inroads and could have the most meaningful long-term impacts.

To gauge the current reach of AI and ML, a business survey was conducted that revealed one in four, or roughly 25,000 to 30,000, San Diego County firms are engaged with these technologies on some level. Ninety-five percent of those companies cited that they had already developed or adopted (or some combination of the two) these technologies, with the other five percent reporting that they plan to develop or adopt AI or ML in the very near future.

Business size is the single most important determinant of AI engagement in San Diego. Firms with more than 50 employees are three to six times as likely to have adopted, developed, or invested in AI technologies compared to those with fewer than 10 employees and two to five times as likely as those with 10 to 49 workers on the payrolls.

This trend is amplified when considering the industry in which a firm operates. Firms in San Diego’s innovation cluster, which includes industries like telecommunications and information technologies, were one and a half to three times as likely to have adopted, developed, or invested in AI technologies compared to population-serving or more traditional industries, such as tourism, healthcare, and financial services. This was true even when controlling for business size.

Drilling down a bit further, three industries emerged as having an above-average proportion of companies engaged with AI and ML:

• Telecommunications and information technologies
• Software and relevant consulting
• Transportation and related manufacturing

Together, these industries—which tend to represent AI developers more than adopters—support an estimated 175,680 local jobs and $33.3 billion in annual GRP, amounting to roughly 10 percent of total employment and 13 percent of GRP. While the number of employees working in this industry cluster is on par with the national average, the average salary in this cluster is $127,960, which is 3.9 percent above the national average for these industries and more than 70 percent above San Diego’s average worker salary of $74,498.

The high pay commanded by workers in these AI-concentrated fields reverberates loudly through the local economy: for every 1,000 jobs gained in this industry group, another 1,400 are created in other sectors.

Despite current economic conditions, three in five employers (62 percent) indicated that they expect to see the number of employees specifically engaged in AI-related work to grow over the next 12 months. These new jobs will provide a sound foundation to support consumer spending throughout the region, benefiting everyone from retailers to bars and restaurants to realtors and construction workers as they come online.

Fields where AI and ML engagement is below average support 497,136 jobs, which is just over 10 percent above the national average for these industries, and $73.8 billion in GRP annually. These tend to represent traditional or population-serving industries, such as healthcare and tourism, where firms are more likely to adopt, rather than develop, AI or ML.

Similar to the trends observed among AI developers, AI adoption is also creating jobs in the region. Sixty-six percent of respondents agreed that the use of these technologies has created new job opportunities at their firms, and 54 percent also noted that using artificial intelligence technologies is increasing the need for more workers at their business. Only 19 percent of employers indicated that AI or ML technology adoption resulted in a reduction in job opportunities.

The positive effects of AI and ML adoption on business activity have already been made apparent. The use of AI or ML technologies largely supports four areas of firm activity: the development of new products and services, improved efficiency and productivity, reduced costs, and an increase in business revenues.

While average salaries and job multipliers are less pronounced, the knock-on impacts from productivity gains in AI-light industries are more than 20 percent greater than those in AI-concentrated industries. Consequently, AI and ML implementation among firms in lower-concentration industries presents a huge business opportunity for AI and ML developers, and the additional output from productivity gains in AI-light industries will have a more marked impact on overall growth in San Diego than it would in AI-concentrated fields.

The pop-out charts below provide some additional context surrounding the spillover effects of employment and productivity on the region as a whole. Toggling between AI-concentrated and AI-light industries shows the relative impacts on the job market and GRP between each set of industries and clusters.

The workforce implications of AI and ML have been widely debated by academics and business leaders alike. Until recently, many labor economists have likened AI proliferation to previous waves of automation, where the lowest-skilled workers were typically automated out of a job.

However, this does not appear to be the case when it comes to AI. A  novel approach  to analyzing which types of jobs could be impacted by AI and ML comes from Stanford PhD candidate Michael Webb. The approach, which was  further explored  by the Brookings Institution, examines AI patent language and extracts the most common verb-object pairs. These verb-object pairs were then compared to job postings in order to develop a picture of which types of jobs could potentially be performed or enhanced by AI.

Applying this approach to job postings in San Diego provides results consistent with the research produced by Webb and Brookings. Two in three local job postings that include verb-object phrases most commonly encountered in AI patents require at least a bachelor’s degree and pay a median salary of $59,300 per year. In other words, these are not lower-skill, lower-paying jobs and reflect a wide variety of positions, ranging from software engineers to retail salespeople to accountants and even truck drivers.

Job postings that include specific language that overlaps with AI patents are becoming more prominent. In September 2016, there were 15,514 unique job postings that included the verb-object pairs above, representing 1.1 percent of total nonfarm employment in San Diego. By July 2019, that number had jumped to 34,273 unique postings, representing 2.3 percent of nonfarm employment. This implies that business demand for the types of skills provided by AI and ML technologies is likely to continue to increase in coming years, presenting another solid business opportunity for local AI and ML developers.

To be clear, saying that a set of occupations or industries could be impacted by AI and ML does not necessarily mean that it will lead to job losses in those fields. As stated, the business survey results revealed that far more firms in San Diego have seen an increase in job opportunities because of AI and ML implementation. This suggests that emerging AI and ML technologies have—at least up to this point—been complementary to the workforce and will not immediately lead to mass layoffs.

As a case in point, Booz Allen Hamilton cites a recent Natural Language Processing engagement where the technology was used to aid help desk technicians connect current tickets with resolved tickets. The organization implementing this solution was not looking to reduce its workforce but increase throughput and offer their employees more meaningful, higher-valued work that could not be completed by automation. 

A closer look at the key AI patent verb-object pairs identified in the job postings data helps to isolate and differentiate in-demand skills sought by hiring managers across the country. It also provides vital insights into whether or not hiring managers are struggling to find qualified workers for certain positions.

Surveyed AI technology developers in San Diego County are largely focused on machine or deep learning, though other technologies and techniques are also being developed locally. Eight in 10 employers indicated that they are involved in the development of machine or deep learning (82 percent); this was the most selected technology. Firms are also engaged in the development of image, speech, or video recognition (46 percent), selected choice natural language processing (46 percent), robotics process automation (24 percent), and autonomous vehicles (21 percent).

Employers in a number of U.S. markets are trying to hire workers in earnest that possess skills readily available from San Diego AI. For example, San Diego-based recognition AI—which survey respondents cited as a primary focus of AI development—could fill the output gaps experienced by companies across the U.S. without leading to job losses. Companies across the U.S. posted the same job posting for recognition-based skills an average of five times and for a duration of 36 days before filling the position or expiring it. This is about 20 percent more time and effort than all other jobs.

San Diego AI technology could therefore lower the costs associated with staffing hard-to-fill positions without disrupting local job markets, since the technology could be employed to fill the gaps left by untimely or unsuccessful recruiting efforts.

The above reflects just one example of how local AI could benefit businesses nationally. Readers can use the interactive map below to obtain additional information on total job postings, unique postings, posting duration, posting intensity, and median salary for each verb-action item for all of the metro areas explored.

Despite AI’s productivity-boosting, job-creating power, a number of challenges remain.

Top of mind for most local employers is the inability to source qualified talent or secure the necessary financing to develop or adopt AI and ML. This was especially true among smaller firms who, as a group, cited lower engagement rates with AI and ML than their larger counterparts.

In particular, San Diego employers are struggling to fill open positions for Network and Computer Systems Administrators, Web Developers and Digital Interface Designers, and Information Security Analysts. Together, there were 2,771 job postings for these occupations between September 2016 and March 2020, yet only 192 people were hired.

That said, COVID-19 and the subsequent increase in remote work has expanded the talent pool for San Diego County’s AI and ML employers. Employers are still facing challenges sourcing qualified individuals, particularly those with relevant industry experience, but the ability to connect with talent outside of the region has been helpful in expanding the qualified applicant pool. While employers always want to have more choice in who they can hire, they were also generally positive about the quality of talent available within San Diego County, and indicated it was a strength of the region. 

Following talent and financing needs, workforce training lands front and center. A majority (57 percent) of respondents said that they had provided training to their workers, but the same percentage also said that additional training would be necessary. As such, AI developers may want to create programs that streamline the training process in order to help companies making the transition to AI and ML adoption. Booz Allen Hamilton can confirm this observation in the study as they have worked with several local employers to help develop and deliver AI training that is aiding in user adoption.  

Surveyed businesses cited job gains from AI and ML implementation, but there’s no guarantee that those new jobs will be distributed across all skill sets and education levels. Employers should prioritize the upskilling of workers in low-skilled positions where AI could become more prominent. This will greatly reduce the odds of job losses while helping to boost incomes.

San Diego is in a strong position to reap a number of benefits from AI.

Contrary to popular belief, AI does not appear to be the job-killing juggernaut that many have feared. The vast majority of surveyed San Diego firms indicated that AI and ML are creating new job opportunities and that interest in filling additional AI roles is high among local employers. The spillover effects will be palpable as these jobs come online.

For companies operating in industries with a lighter AI footprint, the upside potential for productivity gains is promising. Nearly one in 10 survey respondents signaled new engagement with these technologies, which is on par with the past several years. With a steady pipeline of companies looking to transition to AI and ML usage, the share of companies implementing it should grow over the next five to 10 years. The lift to productivity could help boost wages in traditional or population-serving industries and may even have the added benefit of narrowing gender and racial wage gaps.

To that point, employers in San Diego County feel that AI technologies will be both essential and transformational to businesses in the region. This includes firms who have not yet invested in, adopted, or developed AI or ML technologies. If the key barriers to entry, namely, access to qualified talent and financing, are broken down, then the employment and productivity benefits would help create a virtuous cycle of better quality jobs, higher pay, and greater income equality for the region.

All of this is in keeping with  research  that suggests AI and ML technologies will most benefit San Diego (and other markets) if AI is used as a complement to human labor rather than as a substitute. The ability for humans to continually guide AI in executing various work tasks is just as important as companies adjusting their business models to account for what they learn from AI. This form of organizational learning will be pivotal to businesses reaping the greatest benefits from AI and will put San Diego squarely on the vanguard of AI development.

More specifically, San Diego companies and institutions will need to focus on three key items to ensure the region establishes itself as a leader in AI development and adoption:

• Like Booz Allen Hamilton, AI developers should ensure proper workforce training is provided to customers adopting these technologies.
• AI and ML adopters should understand that the returns on investment for AI and ML are largest when these technologies are used to enhance the current workforce, not replace it.
• Colleges and universities should work with local businesses to develop ample and effective academic programs to provide a talent pipeline that is sufficient to meet future AI and ML business demand.

If properly implemented, AI and ML technologies will provide a broad set of benefits to the region. By approaching AI in the ways above, San Diego will be able to achieve and maintain a global reputation as an AI development and implementation hub in the years to come.

Booz Allen Hamilton is honored to serve as the underwriter for Measuring the Future: AI and San Diego’s Economy. With over 1,300 employees in San Diego and as a leader in Artificial Intelligence, Booz Allen Hamilton is at the forefront of AI adoption, development, implementation, and workforce training. We believe San Diego companies can lead this capability globally - attracting talent, driving economic growth, and fueling local job creation in our region. One of our biggest takeaways from this report was that local respondents indicated that AI is truly helping the San Diego economy by creating more jobs rather than eliminating them. People are essential to the ethical application of AI and this technology will empower organizations and – importantly – their workforce to increase productivity, quality, and efficiency. We hope you gained valuable information from this report on the benefits of AI adoption—and what it means for San Diego, and other regions in the future.

This study was made possible by financial contributions from the following sponsors, along with the leadership and guidance of the individuals who participated in the study's Advisory Committee.