Does automation from artificial intelligence (AI) replace workers or augment their productivity? This question elicits contrasting perspectives. According to Frank et al., (2019, p. 6531), barriers that impede researcher’s ability to effectively measure the impact of AI on future of work and thereby, better analyze the question, includes “lack of high-quality data about the nature of work, lack of empirically informed models of key microlevel processes and insufficient understanding of how cognitive technologies interact with broader economic dynamics and institutional mechanisms”. In their paper, Frank et al., (2019) discuss these barriers and provide potential solutions to reduce them (see Table 1 below for overview).
The first barrier refers to the sparse amount of data on skills. Due to this lack of data, current methods for measuring skills are imperfect. One example mentioned by the researchers describes measuring skills based on income; however, this measurement is limited by the fact that it only distinguishes between low-, medium- and high-skilled jobs rather than the actual skills required for a job. Improvements in data collection, specifically improving the ability to collect more specific data for job skills, could clarify labor trends and thereby improve scientist’s forecasting of the effect of automation and AI on the labor market.
Secondly, Frank et al., (2019) discuss the limited modeling of resilience, which refers to understanding labor’s resilience to technological change. To understand the features of a labor market that leads to resilience, one must understand the dynamics of “microscopic workplace skills in combination to produce macroscopic labor trends” (Frank et al., 2019, p. 6535). Two of the solutions mentioned for the limited modeling of resilience include (i) mapping skill interdependencies to inform which types of jobs would be augmented or replaced with new technology and (ii) connecting worker mobility to different skills in order to analyze this relationship and how it relates to job opportunities.
Lastly, the researchers discuss places in isolation and how technological disruptions have uneven effects across geography. For example, cities tend to experience technological change more often than other areas. This difference has implications for the difference between rural and urban areas in terms of the labor force and income distribution. Technological disruptions alter demand for certain skills and, as previously discussed, a lack of data on the actual skills for particular jobs, further masks the differences in the workforce across geographies. The accompanying example employed by Frank et al., (2019) describes how a job-seeking software developer would most likely need to list specific skill sets when applying for a job in Silicon Valley versus other, potentially more rural, areas. Solutions for this uneven effect of technological change across geographies, as stated by the researchers, include improving models for spatial interdependencies with more “granular skills data and new insights into the mechanisms that create today’s cross-sectional geographic trends” (Frank et al., 2019, p. 6537).
(Frank et al., 2019, p. 6534)
Works Cited