AI is creating super-companies while others fall behind

The fact that skills shortages exist in the tech industry is well known, with nearly all UK businesses reporting an IT skills gap in 2023. Such reports have become commonplace in recent years, with the authors arguing, quite rightly, that a lack of skills can undermine the development and adoption of new technologies.

Indeed, in her recent book Essential, Upwork's Kelly Monahan highlights how the skills shortage in the US is likely to equate to around 85 million, costing the American economy up to $8.5 trillion by 2030. It's a phenomenon caused in part by a lack of investment in workplace training, but also by demographic changes, a decline in workforce participation, and challenges involved in recruiting from external talent pools.

What is perhaps less well understood is the role a lack of skills plays in the adoption of new technologies across economies. That's where a recent study from the University of Warwick comes in, as they looked to understand how the level of skills among the general public affects both the breadth and depth of adoption of technologies, such as generative AI.

Rates of adoption

The study uses UK data to examine the spread of two distinct waves of technology: the personal computer boom in the 2000s (the "first wave") and the more recent rise of cloud computing and AI in the 2010s (the "second wave"). The findings show that different types of skills – general degree-level skills and specialized STEM skills – play a key role in driving these adoption patterns.

In both waves, there are clear links between technology adoption and the availability of general and STEM skills at the regional level. To study the first wave of PC adoption, the researchers used data from Harte-Hanks, a marketing firm with deep roots in the ICT supply sector. For the second wave, they analyzed job ad data from Lightcast to track the uptake of newer technologies.

When considering other factors – like population density, industry mix, unemployment, and even London-specific effects – STEM skills show a unique advantage. They are more closely linked to the adoption of second-wave technologies like AI and machine learning than general degree-level skills.

Firm differences

The researchers found that there were clear advantages for companies having STEM-related skills when it came to the adoption of technologies like AI and cloud computing. What's more, the more STEM-related skills were present, the less important high-level skills in other fields became.

The study shows that specific computing skills were important for the adoption of cloud technologies, but more general scientific skills seem to matter more for AI adoption. Interestingly, soft skills were found to have a limited impact on adoption.

This manifested itself in adoption rates almost three times greater among organizations with higher levels of STEM skills. Indeed, this was even evident at a geographic level, with regions with higher levels of STEM skills adopting new technologies at a similar rate to individual firms.

At the organizational level, there was a clear discrepancy evident, with a relatively small number of companies dominating the adoption of new technologies. Indeed, the study found that just 20 companies accounted for around 60% of all job vacancies in AI-related fields. While this has fallen a bit in recent years, the concentration remains extremely high, with the top 100 firms accounting for 80% of all tech-related vacancies.

“The data shows that technologies like AI and cloud computing are widening the productivity gaps between firms and areas with high STEM skills and their lower-skilled peers. The technologies aren't leveling up but rather making differences starker.”

Uneven distribution

The researchers believe, quite rightly, that this uneven rate of distribution presents various challenges, both from a productivity and equality perspective. The data correlates tightly with general productivity differences across the UK, with the new technologies often making things worse rather than better.

The data shows that technologies like AI and cloud computing are widening the productivity gaps between firms and areas with high STEM skills and their lower-skilled peers. The technologies aren't leveling up but rather making differences starker.

This is largely because of the growing concentration of technologies like AI among a small and select number of companies and regions, which ensures that economic growth is far from uniform across the country.

Closing the gap

The study reminds us that there is a clear link between skills and technological progress, with the skills gap playing a clear role in the uneven economic growth we're seeing in so many countries today.

It suggests that companies, and indeed regions, with strong STEM skill sets will benefit from generative AI and similar technologies, but the same cannot be said for companies and regions without these skills. The authors suggest that without concerted action, these disparities are likely to deepen, with high-tech productivity gains concentrated within a narrow set of firms and regions rather than dispersed evenly across the economy.

They urge all stakeholders to take a more proactive and strategic approach to skills development, whether through incentivizing workforce training, investing in STEM education, or forging better links between employers and schools.

Only by expanding access to essential STEM skills across the workforce can policymakers hope to ensure that the transformative potential of emerging technologies becomes a driver of broader economic prosperity – rather than an agent of division.