Fast food workers

We use the Standard Occupation Classification 5 or 6 digit demographic information from the 2020 CPS, available from https://www.bls.gov/cps/cpsaat11.htm and https://www.bls.gov/cps/cpsaat11b.htm. If there is no demographic information (omitted because of data limitations/privacy information), we use the demographic characteristics from the Minor Group (SOC3) in place of the SOC5 demographic characteristics. To calculate the weighted average, we take the employment available from the CPS and aggregate to the job classification level. For Sales Engineers, we use Sales and related workers, all other. For makeup artists, we use Personal care and service occupations.

We use the Standard Occupation Classification 5 or 6 digit demographic information from the 2020 CPS, available from https://www.bls.gov/cps/cpsaat11.htm and https://www.bls.gov/cps/cpsaat11b.htm. If there is no demographic information (omitted because of data limitations/privacy information), we use the demographic characteristics from the Minor Group (SOC3) in place of the SOC5 demographic characteristics. To calculate the weighted average, we take the employment available from the CPS and aggregate to the job classification level. For Sales Engineers, we use Sales and related workers, all other. For makeup artists, we use Personal care and service occupations.

Source: US Department of Labor

Available from https://www.onetonline.org/

State projections are developed in the labor market information sections of each State Employment Security Agency (SESA). The projection period is 2018-2028, which includes the long-term period calculated up until 2028.

Source: US Department of Labor

A score of 5 represents the highest possible exposure to machine learning (ML), while a score of 1 represents little exposure to ML at all.

This is the Suitability for Machine Learning (SML) score available in Brynjolfsson, Mitchell, and Rock (2018) and Brynjolfsson, Frank, Mitchell, Rahwan, and Rock (2022). The aggregate score reflects a 21 question rubric designed to measure the overall exposure of a given Detailed Work Activity (DWA) available from the O*NET database. Activities are aggregated with equal weight within task, and tasks are aggregated using importance weights into occupations following the O*NET crosswalks between DWAs, tasks, and occupations. This is not an automation measure, but rather represents the relative extent to which an occupation will be impacted in some manner by machine learning and artificial intelligence.

Source: Brynjolfsson, Mitchell, and Rock (2018) & Brynjolfsson & Frank, Mitchell, Rahwan, and Rock (2022)

A score of 5 indicates high data intensity in the job, and a score of 1 indicates low data intensity for that job category.

This is an average of a subset of SML rubric items designed to represent the quantity of available data or exposure to data in this particular occupation.

Source: Brynjolfsson, Mitchell, and Rock (2018) & Brynjolfsson & Frank, Mitchell, Rahwan, and Rock (2022)

This is the standardized (mean zero, standard deviation 1) score representing the routine cognitive work component of the job category. These scores come from Acemoglu and Autor (2011) and represent the extent to which the job category contains repeated tasks of a cognitive variety (as opposed to non-routine or abstract cognitive tasks or manually performed physical tasks). These scores are constructed using O*NET categories.

The Remote Work Potential Score that represents whether a job can be performed remotely from Dingel and Neiman (2020).

A score of 1 indicates a likely remotable occupation, and a score of 0 indicates an occupation with no remote work potential. Scores between 0 and 1 represent the aggregate of different constituent jobs that belong to the broader category indicated in the table.

Source: Dingel and Neiman (2020)