Do we want less automation?

• 成果类型: Editorial Material
• 署名作者: Agrawal, Ajay; Gans, Joshua S.; Goldfarb, Avi
• 署名单位: University of Toronto
• 刊物名称: SCIENCE
• ISSN/ISSBN: 0036-9672
• DOI: 10.1126/science.adh9429
• 发表日期: 2023-07-14
• 页码: 155-158
• 关键词: inequality

摘要

Impressive achievements made through artificial intelligence (AI) innovations in automating the tasks required in many jobs have reinforced concerns about labor market disruption and increased income inequality. This has motivated calls for change in the direction of AI innovation from being guided by task automation to instead focusing on labor augmentation (1). But task automation and labor augmentation are not polar opposites. Instead, automation of some tasks can lead to augmentation of labor elsewhere. Furthermore, AI automation may provide a path to reversing the trend of increasing income inequality by enabling disproportionate productivity improvements for lower-wage workers, allowing them to perform at levels that would previously require years of education and experience.