Automation and the Workforce: A Firm-Level View from the 2019 Annual Business Survey

Automation and Employment Over the Technology Life Cycle: Evidence from European Regions

“…Overall, the adoption of AI among firms remains relatively low. Between 2016 and 2018, the percentage of firms using or testing AI in the U.S. was reported to be 3.2% (Acemoglu et al 2022). Furthermore, research indicates that adoption tends to be more prevalent among larger and older firms (Zolas et al 2021, Acemoglu et al 2022).…”

Reshoring, Automation, and Labor Markets Under Trade Uncertainty

“…We calibrate the scale of the automation cost function to ν = 8.57 such that the model implies a steady-state automation probability of q a = 0.096, or about 38 percent at the annual frequency, which lies within the range of firm-level estimates. For example, in a recent study based on the 2019 Annual Business Survey (ABS) of the U.S. Census Bureau, Acemoglu et al (2022) report that, in total, 30.4 percent of U.S. workers are employed at firms using advanced technologies for automating tasks. Exposure to automation is higher in manufacturing, with 52 percent of manufacturing workers employed at firms using advanced technologies for automation.…”

Discriminatory bias (and how to prevent it) in the European Union: a review of the ethical and regulatory framework for artificial intelligence.

“…To achieve this, the following content structure is proposed: in the first section, the concept of algorithmic discrimination will be introduced from a multidisciplinary perspective; in the second section, the main results of the quantitative and qualitative systematic review of the approach to the issue of discriminatory bias in the main European regulatory instruments and recommendations related to the design, development, implementation and use of AI systems will be presented; and finally, a third section will aim at systematising the recommendations to minimise and mitigate this risk. In short, this proposal makes it possible 2 The AI risks that have raised the most concern include the following: 1) AI algorithms can perpetuate and amplify existing biases in the data, leading to discriminatory outcomes (bias and discrimination) (Mayson, S, 2019); 2) many AI models, especially the more advanced ones, are 'black boxes' that provide little or no insight into how they reach their conclusions (lack of transparency) (Molnar, 2022;Ribeiro et al, 2016); 3) the use of personal data in AI raises concerns about privacy and consent (ethical and privacy issues) (Richards, 2021;Véliz, 2021); 4) data quality is critical to AI performance, and faulty data can lead to erroneous results (data quality dependency) (Byabazaire et al, 2020); 5) AI-driven automation can displace human jobs, creating economic and social challenges (unemployment and job displacement) (Acemoglu, et al, 2022;Acemoglu & Restrepo, 2019;Frey & Osborne, 2017) 6) AI can be used for harmful purposes, and AI systems are vulnerable to attack and manipulation (security and misuse) (Brundage et al, 2018) .…”