Are Robots Really Stealing Our Jobs? Perception versus Experience

Abstract

Popular narratives warn that we should beware of robots replacing employees in the workplace. However, prior research has been limited by data availability. Studies typically use data based on geographic locations and aggregated industrial robot and employment variables and do not directly measure whether someone has lost their job to a robot. In this data visualization, the author presents results from a national survey that includes questions that address this issue. The results reveal that almost 14 percent of respondents in the sample report having lost their jobs because of employers replacing their jobs with robots. The results also show that respondents’ perceptions of the percentage of employees who have been replaced by robots is inflated. The difference between respondents’ perceptions and their own experience is exacerbated for those who report having lost jobs to robots.

Popular media outlets as well as social media feeds often proclaim that we should be wary that robots are standing ready to replace employees in the workplace. Results from a limited number of studies on this topic, however, are mixed. For instance, Acemoglu and Restrepo (2020) found that the incidence of robots has a negative effect on employment, while Autor and Salomons (2018) observed the opposite effect. Findings from my study published in Socius (Dahlin 2019) confirm the analysis of Autor and Salomons. I argue that the positive association between robots and employment can be explained by complementarity that exists between employees who possess the appropriate skills and expertise to collaborate with robots in the workplace. As an example, on assembly lines, employees may work alongside robots that restock workstations or retrieve parts. Or computer scientists and engineers may develop, program, use, or maintain robots or work with them to perform technologically advanced responsibilities.

Much of the research on this topic has been limited by issues related to data availability. Robot and employment data are typically aggregated at the national (e.g., Klenert, Fernandez-Macias, and Antón forthcoming) or subnational (e.g., Acemoglu and Restrepo 2020) level and do not directly measure whether employees have lost their jobs to robots. Instead, they measure the copresence of robots and jobs in geographic locales. These data may not provide a complete picture, because even when robots are displacing some types of jobs in these locations, losses may be masked by overall gains in other sectors of employment. Extant scholarship is also limited in the types of robots that are studied. The most readily available variables measure industrial robots, which the International Federation of Robotics (2017:32) defines as “automatically controlled, reprogrammable” machines for use in industrial settings. But other types include robots that clean, carry luggage at airports, care for the needs of the elderly, and search for books in libraries.

I present results in this data visualization that address these limitations. The results are based on a national survey (n = 1,959) that I commissioned Qualtrics to conduct and come from two survey questions. The first question asked respondents about their perceptions about jobs’ being replaced by robots. This question asked respondents to estimate the percentage of employees whose employers have replaced their jobs with robots. The second question asked respondents whether employers have ever replaced their own jobs with robots.

Figure 1 highlights respondents’ perceptions of the extent to which employees’ jobs have been replaced by robots. The bars in the figure disaggregate respondents’ perceptions by whether they have ever lost jobs to robots. Respondents who reported they have not had jobs replaced by robots estimated that 29 percent of employees have had their jobs replaced. Respondents who reported that they have had jobs replaced by robots estimated that 47 percent of employees’ jobs have been replaced. The dashed line indicates that almost 14 percent (13.7 percent) of respondents reported that their jobs have been replaced with robots.

Figure 1.

Respondents’ perceptions of the percentage of people who have lost jobs to robots. The figure shows a sizable difference between respondents’ perceptions on the basis of whether their jobs have not been replaced by robots compared with respondents whose jobs have been replaced (29 percent compared with 47 percent). Respondents’ perceptions vastly overestimated those in the sample who reported experiencing jobs’ being replaced by robots (13.7 percent). The data come from a national survey conducted in the United States by Qualtrics in September 2021 (n = 1,959).

The results have two notable implications. First, respondents’ perceptions of others’ losing their jobs to robots do not match the experiences of those in the sample. For respondents who did not lose jobs, a twofold increase exists between their perceptions of jobs lost to robots compared with the results for the entire sample (29.0 percent compared with 13.7 percent). For those who did report losing jobs, more than a threefold increase exists between their perceptions and the results for the entire sample (46.9 percent compared with 13.7 percent). Respondents’ perceptions are exaggerated compared with (and no doubt influenced by) the attention-grabbing headlines predicting a dire future of employment. Second, a comparison of perceptions of others by whether respondents lost jobs to robots indicate that one’s own job experience biases one’s perception that others have had the same experience.

Supplemental Material

sj-docx-1-srd-10.1177_23780231221131377 – Supplemental material for Are Robots Really Stealing Our Jobs? Perception versus Experience

Supplemental material, sj-docx-1-srd-10.1177_23780231221131377 for Are Robots Really Stealing Our Jobs? Perception versus Experience by Eric Dahlin in Socius