The people dilemma: How human capital is driving or constraining the achievement of national AI strategies

College classroom with math on chalkboard

In the early days of the COVID-19 pandemic (June 2020), LinkedIn released a report showing that the demand for AI skills had cooled down—but by October 2020, demand had already come roaring back. This is not surprising: according to the 2020 RELX Emerging Tech Executive Report, AI adoption soared during the pandemic, and a staggering 68% of companies increased their AI investment during the year. Further, 81% of companies now report using AI technologies, up 33 percentage points since 2018.

Companies are increasingly using AI technologies on mission-critical applications, which has led to an explosion in the need for data scientists and technologists to build and support these applications. Not surprisingly, 39% of companies now cite a lack of technology expertise as a leading stumbling block to AI usage and adoption.

Despite the value of machine learning, much of AI development is still predicated on two pillars: technologies and human capital availability. Our prior reports for Brookings, “How different countries view artificial intelligence” and “Analyzing artificial intelligence plans in 34 countries,” detailed how countries are approaching national AI plans, and how to interpret those plans. In our most recent follow-up piece, “Winners and losers in the fulfillment of national artificial intelligence aspirations,” we discussed how different countries were fulfilling their aspirations along technology-oriented and people-oriented dimensions. In this, our first follow-up analysis, we dive more deeply into the people dimension of our typology, paying close attention to skills gap and attainment.

Development of human capital factors

We followed the same strategy in examining the people dimension of AI strategic plan fulfillment as we did with our prior post. In this case, we had three data elements that comprised our people dimension: Relative Skill Penetration (the prevalence of AI skills for the average occupation in the country), AI Hiring Index (the percentage of LinkedIn members within a given country that had AI skills reflect in their profile), and STEM graduates (the number of graduates with STEM degrees in any given country). The first two data elements were from Stanford’s Human Center Artificial Intelligence work while the STEM graduate information was from the World Bank.

Rather than attempting to interpret three data elements in isolation from each other, we conducted a factor analysis to determine if any of the three data elements were closely related. Closely related items were mathematically combined into a single composite factor which contains both data elements to aid in the interpretation of the data.

Our factor analysis showed Relative Skill Penetration and AI Hiring Index were closely related to each other and formed a single composite factor. Our other data element, STEM graduates, was not mathematically related to the other data elements, and so our interpretation is based on these two factors. The first factor reflects the current job market for AI, since it is based on professionals and job postings that currently exist.

The second factor is based around STEM students within each country, which reflects future additions to the job market. As these STEM students graduate, they will change the state of the job market. Using the two factors, we can interpret the people dimension within two distinct sub-dimensions: Present Market and Future Market.

Figure 1 shows where a select group of countries sit along these sub-dimensions.

Chart of the present and future state of the people dimension of AI strategy

We interpret and name the quadrants as follows. The countries that are in the upper right-hand corner we dub “Leaders”; they have both a robust current market (factor 1) and a strong incoming supply of qualified STEM students (factor 2). Countries in the lower right quadrant we dub “Future Prepared” and these are countries that have an incoming supply of qualified STEM students, but their current job market is weaker. The countries in the upper left quadrant – we dub the “Present Prepared” and are those countries that have a robust current job market, but they lack a strong supply of incoming talent. Finally, the lower left quadrant – we dub the “Unprepared” quadrant and these reflect countries that neither have a robust current job market, nor do they have a strong supply of incoming talent.

India, Germany, and Singapore

India (94th percentile in current market and 92nd percentile in future market), Germany (73rd percentile in current market and 98th percentile in future market), and Singapore (82nd percentile in current job market and 94th percentile in future market) are all ideally positioned for meet the human capital demands of AI work. All three countries have a robust current market of technically skilled people, and are positioned to generate even more with their current educational path. We see no people-centric issues that are likely to hinder them.

China and South Korea

Both China (48th percentile in current market and 96th percentile in future market) and South Korea (50th percentile in current market and 86th percentile in future market) are in the same position. While they do not currently have a robust market for AI technologists, they are pushing an impressive number of college students into the STEM fields; over time, this will shift them into the Leaders quadrant. For these countries, the question is how quickly they can convert their incoming swell of STEM graduates into the job market, and how many—if any—of their graduates leave for other countries.

US, Canada, Australia, and Sweden

All four of these countries face a similar problem. While they currently have a strong market for skilled technologists, they are about to drop off a cliff without a strong incoming set of STEM-skilled students. While Canada (80th percentile in current market and 38th percentile in future market) is in a marginally better position than the other three countries, all four countries need to push harder to encourage more residents to enter STEM fields. For the US (80th percentile in current market and 18th percentile in future market), this problem is likely to get worse: the majority of their current STEM students are from other countries and are likely to return to their home countries upon graduation. As such, we believe that the US position may be artificially optimistic, since our data does not distinguish the citizenship of current STEM students within the country.

The U.S. ranks poorly

At present, India, Singapore, and Germany are in a strong position and are also developing sufficient human capital to achieve their national AI strategies. All our analyses indicate that their current markets are good, and their future markets are also strong. China and South Korea are playing catch-up to India, Singapore, and Germany, as their current job market is weak but they have a very strong base of incoming talent.

There is no way around the bad news for the US, Canada, Australia (62nd percentile in current market and 20th percentile in future market), and Sweden (48th percentile in current market and 72nd percentile in future market). All four countries are going to be in trouble with a lack of AI talent unless they immediately and dramatically improve the number of STEM graduates in their colleges and universities. While they are surviving based on the number and talent that currently exists, the future appears bleak unless strong measures are taken.

In our next post, we will dive more deeply into the technology dimension of the fulfillment of national AI strategic plans. But, unmistakably, these findings are a clarion call to the U.S. to make dramatic changes in STEM education now or be soon relegated to second-tier status.