1. Shared Prosperity in the Era of AI
The 7th World Humanities Forum
Busan, Korea
November 8-10, 2023
Kang-Kook Lee
Ritsumeikan University
2. Challenges in the Coming AI Era
• Rapid automation along with the development of AI
(Artificial Intelligence)
• Co-existence with machines leading to increasing
productivity
• Examining the impact of technological progress such as
AI on the labor market
• The possibility of a ‘robocalypse’, and concerns about
the inequality effect of technology
• What is to be done to promote shared prosperity and
inclusive growth in the AI era?
3. Automation Leading to Mass
Unemployment?
• The concern about the robocalypse was always there in history since
the Luddites
• For example, a speech by President Kennedy in 1962
• Recent studies after the development of automation technology report
mixed results about replacement of human labor by machines
• Frey and Osborne (2013) and other studies report a large risk but
other studies’ estimation is lower because one occupation does
several tasks (Arntz et al., 2016, Nedelkoska and Quintini, 2018)
• Profit motivation, political resistance and government policy should be
considered
• Historically, more and new jobs were created with higher
productivity,(60% of jobs in 2018 didn’t exist in 1940) so the idea of
the about the robocalypse is not correct (Autor et al., 2022)
6. Automation and Rising Inequality
• A valid concern about the effect of technological progress on rising
inequality
• The old model about the race between technology and education with
skill-biased technological change, resulting in the education premium
• A task-based study reports a polarization of the labor market because
routine jobs with the middle skills could be easily automated,
compared to low and high-skilled jobs
• So a U-shaped relation between the skill level and employment
change, after the 1990s (Autor and Dorn, 2013)
• However, wages in lower-skill occupations could also decline was
workers with middle-skill levels could enter lower-paid service
occupation, so the U-shaped pattern of wage growth is less certain
7. A U-shaped relationship between the skill level and change in
the employment, reported by the task-based approach
8. Empirical Study on
Automation and Inequality
• Empirical evidence for the positive effect of automation on overall wage
inequality
• Workers exposed to routine jobs in industries with more automation saw a
stagnation in wages after 1980, and the task displacement explained more
than 50% of the rise in wage inequality (Acemoglu and Restrepo, 2022)
• Also, more industrial robots lead to lower wage and employment
(Acemoglu and Restrepo, 2020)
• Technological advancements measured by patents are associated with
negative labor market outcomes (Kogan et al., 2021)
• But workers’ bargaining power matters to the change in the labor share
(Stansburry and Summers, 2020)
• Proper government policies and political changes are necessary for
shared prosperity with the rapid technological change and automation
(Acemoglu and Johnson, 2023)
9. More task displacement due to automation is associated
with the lower wage growth after 1980
10. AI, Unemployment and Inequality
• AI is different from former automation because it can handle tacit
knowledge, and can replace non-routine labor?
• Many establishments have introduced AI, and demanding less of non-AI
jobs employment, but the aggregate or industry level results are unclear
• Goldman Sachs reports 25% of current work tasks could be automated by
AI, but 7% of jobs are at significant risk of automation because of AI, and
most jobs will be augmented by AI (Hatzius et al., 2023)
• Limitations in the replacement of labor by AI, because of comparative
advantage of humans in creativity, judgment, emotional intelligence, etc.,
and costs of robots doing dexterous and adaptive interactions
• The inequality effects of AI: several studies report the generative AI like
ChatGPT reduces inequality in the same occupation, by encouraging
productivity of less-skilled workers more, but aggregate effects may be
negative to distribution (Brynjolfsson et al., 2023; Noy and Chang, 2023)
11. Big concerns about AI but limited possibility to
replace labor in reality
12. Studies on the effect of generative AI report more
benefits for the lower-skilled workers in customer
service and writing.
13. AI and Prosperity for All
• How can we maximize the benefit of AI, while minimizing its risk such as
the inequality effect?
• Many advocate for complementary institutional and policy reforms: public
investment in education and training, protecting and empowering workers,
effective regulation of big techs and platform companies, etc. (Autor, 2022)
• Excessive automation without substantial productivity effects and
reinstatement effects could occur (Acemoglu, 2021)
• Efforts to change the course of AI development to foster new labor-
intensive tasks and promote productivity, rather than merely cutting labor
costs are necessary, such as in education and healthcare
• Political changes for this transformation based on stronger power of
workers and a proper vision, leading to a larger role of the state
• As Marx said, the problem is not machinery in itself but how we use it!