By Stijn Broecke.
Last month, chatbot Rose won the 2015 Loebner Prize for artificial intelligence (AI) – an annual contest in which machines try to fool judges into believing that they are human. While Rose ranked the most human-like bot by three of four judges, it failed to fool any of them into thinking it was a real person.
While this may be disappointing news for those working in the world of AI, it is good news for those who are worried that robots will one day steal our jobs. Indeed, while some analysts believe that half of our jobs are at risk of being computerised over the next two decades (Frey and Osborne, 2013), others are much less pessimistic (Boning, Gregory and Zierahn, 2015). The fact remains that robots have persistently failed to imitate the most human of skills, such empathy, teamwork, relationship building, etc.
But this is also why, according to a recently published NBER working paper by David Deming, social skills (defined as the ability to work with others) have been growing in importance in the labour market. While technology may be reducing the demand for some routine skills, it is simultaneously increasing the demand for more difficult-to-automate social skills.
In this thought-provoking paper, Deming documents a number of interesting trends in the United States labour market. The first of these is that tasks performed by workers have increasingly required social skills. Between 1980 and 2012, social skill task inputs grew by 24%, compared to just 11% for non-routine analytical tasks (Figure 1). At the same time, there has been a strong decline in routine task inputs over the period. Interestingly, Deming finds a growing complementarity between cognitive skills and social skills: since 1980, employment growth has been particularly strong in occupations that require both high cognitive and social skill.
In line with the above, real wage growth since 1980 has been greatest in occupations that require workers to have strong social skills (Figure 2). In comparison, wage growth for jobs with high mathematics and low social skill requirements has been positive but relatively modest, while real wages have declined for nearly all jobs that are below the median in both mathematics skills and social skills.
According to Deming, these trends may also explain some other developments that have taken place in the labour market, like the narrowing of gender gaps in labour market outcomes. For men, the task content of work has barely changed since 1980, while for women there has been a dramatic decline in routine task intensity, matched by an increase in social skill task inputs.
Of course, Deming’s work fits into a wider economics literature which has shown that there is a broader set of non-cognitive skills, such as perseverance, self-control, openness to experience, the capacity to work collaboratively or as part of a team, communication skills, etc… which are also of importance in the modern workplace.
Such research echoes the view of employers around the world: when asked why they have difficulty filling jobs, 34% of employers cite a lack of technical competencies, but 17% also cite a lack of “soft skills” (ManpowerGroup, 2015).
While social skills may be important in the labour market, the elephant in the room is obviously what policy can do to improve the social skills of workers – particularly of those who are being left behind by technological change. This is an area where knowledge about what works is somewhat lacking – although evidence is gradually emerging, for instance about interventions that successfully improve the non-cognitive skills of youth at risk of labour market exclusion (Carcillo et al, 2015). Given the trends documented in Deming’s paper, this is clearly an area where more research will be needed in years to come.