This site is undergoing review. Some annotations were human-generated, some AI-generated — all are being verified.

The Economics of Artificial Intelligence: Implications for the Future of Work

Ernst, Merola, Samaan

2019IZA Journal of Labor Policy376 citations

Review / survey / meta-analysisTheoretical model

AI (General)Automation / RobotsGeneral automationDeveloping economiesAugmentation vs. substitutionTraining / upskillingPlatforms / gig economyAlgorithmic management

DOI: 10.2478/izajolp-2019-0004

Abstract

Abstract The current wave of technological change based on advancements in artificial intelligence (AI) has created widespread fear of job loss and further rises in inequality. This paper discusses the rationale for these fears, highlighting the specific nature of AI and comparing previous waves of automation and robotization with the current advancements made possible by a widespread adoption of AI. It argues that large opportunities in terms of increases in productivity can ensue, including for developing countries, given the vastly reduced costs of capital that some applications have demonstrated and the potential for productivity increases, especially among the low skilled. At the same time, risks in the form of further increases in inequality need to be addressed if the benefits from AI-based technological progress are to be broadly shared. For this, skills policies are necessary but not sufficient. In addition, new forms of regulating the digital economy are called for that prevent further rises in market concentration, ensure proper data protection and privacy, and help share the benefits of productivity growth through the combination of profit sharing, (digital) capital taxation, and a reduction in working time. The paper calls for a moderately optimistic outlook on the opportunities and risks from AI, provided that policymakers and social partners take the particular characteristics of these new technologies into account.

Summary

Ernst, Merola, and Samaan synthesize existing literature on automation and robotization to develop a conceptual framework for understanding AI's implications for employment, productivity, and inequality, arguing that AI differs from previous waves through its focus on mental tasks and potential to augment low-skilled workers.

Main Finding

AI holds potential for productivity gains especially among low-skilled workers and in developing countries due to reduced capital costs, but risks increasing inequality through first-mover advantages and market concentration unless policies ensure competition, data protection, and broad sharing of technological rents through skills policies, taxation, profit sharing, and working time reduction.

Primary Datasets

Penn World Tables version 9; ILO Trends Econometric Models

Secondary Datasets

McKinsey Global AI

AI-focused

ILO statistics on consumer spending; ICF Global Coaching Study; patent data from Fujii and Managi (2018); McKinsey Global Institute reports

Key Methods: Conceptual analysis and literature review integrating task-based frameworks, historical analysis of automation waves, and policy synthesis
Sample Period: Historical perspective from 19th century to 2030 (projections); descriptive data 1950-2017
Geographic Coverage: Global perspective with focus on advanced economies (G7) and developing countries
Sample Size: G7 countries for productivity analysis; global coverage for job polarization trends
Level of Analysis: Individual, Firm, Occupation, Industry, Country, Task
Occupation Classification: None
Industry Classification: None

Notes

IZA Journal of Labor Policy, vol. 9, no. 1, pp. 1-35 [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves. [Claude classification]: This is primarily a conceptual/review paper rather than an original empirical study. The authors synthesize existing literature on robotization and automation to draw implications for AI. The paper includes some descriptive statistics from secondary sources (e.g., productivity trends from Penn World Tables, job polarization trends from ILO models) but does not conduct original empirical analysis. The theoretical model draws on task-based frameworks and endogenous technical change theory to analyze how AI differs from previous automation waves.