Historically, technological changes have created more jobs than they destroyed; but this time could be different.
Acemoglu and Pascual Restrepo PhD ’16 conducted research analyzing automation data across 49 industries between 1987 and 2016 using AI technology. They discovered that economic activity associated with AI correlates to increased income inequality for multiple reasons, including technological developments that make AI accessible at lower costs or automation’s commoditizing effect on employment levels.
Job creation
New research indicates that AI’s growth has raised concerns over job loss due to automation; however, new evidence indicates these fears are exaggerated. A shift towards automation may even create more jobs than it destroys; key is making sure those displaced by AI find employment in occupations which are expanding or creating entirely new positions.
Goldman Sachs anticipates that AI will lead to the creation of hybrid occupations that combine human and machine work, and displaced workers will be retrained for these occupations. Furthermore, technology-induced job losses have historically been offset by broad-based gains in aggregate real incomes which increase demand for products and services.
Researchers anticipate the greatest effect of AI technology will be felt in industries with repetitive, routine activities susceptible to automation such as retail, paralegal work and data processing tasks. They include occupations like retail salesperson, paralegal worker and data entry operators.
Job destruction
As AI becomes more widely utilized, there is growing fear that it will displace numerous jobs. The fear is founded in automation’s ability to replace humans more efficiently than they could ever manage – leading to massive job losses and economic opportunities lost forever.
Automation may seem threatening to jobs; but in reality it will likely transform them rather than eliminate them, according to research by Daron Acemoglu and Pascual Restrepo PhD ’16 from MIT. Their study concluded that technological substitution has far less of an impact than generally predicted across most occupations.
Researchers examined data on machinery and software use across 49 industries, as well as occupations and wages in each sector. On average, when jobs were lost due to automation they were replaced by equal numbers of new jobs being created in that same industry – in line with Schumpeterian creative destruction theory; nonetheless, their findings do not diminish those whose jobs have been taken over by technology.
Wage increases
Automation may bring job loss, but it can also increase their pay. Furthermore, automation can assist workers by improving worker productivity – for instance by using AI to communicate more efficiently with customers via email – leading to higher wages for workers such as landscape gardeners whose jobs have been automated.
However, AI does not always benefit everyone equally; for some workers increased wages may be offset by decreased work hours or overall demand resulting in some being worse off than they would have been without its implementation and increasing inequality over time.
Acemoglu asserts that companies should see AI as a complement to human work rather than as an attempt to replace it, so as not to undercut workers’ pay and their earnings potential. Furthermore, workplace organization and policy advocacy may help workers maintain stable compensation structures and increase earning potential.
Inequality
Most economists agree that AI won’t cause unemployment to increase in high-income countries; rather, it may increase inequality by shifting income between low, medium, and high skill jobs, decreasing labor’s share in overall real income, as well as possibly decreasing demand and weakening unions.
One study shows that AI investment leads to increased incomes among top decile households while decreasing those of bottom two deciles, due to lower labor costs and greater productivity associated with AI investments. It also increases real income generated from exports of modern services tied to such investments.
Though automation may cause widespread unemployment, evidence contradicts these concerns. Other forces have contributed to rising inequality and stagnant wage growth such as hollowing-out of manufacturing sector economies and shift toward service-oriented economies.
