How Long Before AI Starts Replacing Human Workers?
Longer Than You Think, New Study Says. A study from MIT suggests businesses should expect a 'gradual' integration of artificial intelligence into their operations.
BY CHLOE AIELLO, REPORTER@CHLOBO_ILO
While artificial intelligence is expected to augment productivity in businesses across industries, it's also seen as a significant threat, as many U.S. workers worry that their jobs could soon be given to AI programs.
A new study from the Massachusetts Institute of Technology found that there are still relatively few tasks that companies can afford to replace with AI, however, meaning the days of AI replacing many human jobs may still be a ways off.
"This indicates a more gradual integration of AI into various sectors, contrasting with the often hypothesized rapid AI-driven job displacement," Neil Thompson, director of MIT's FutureTech research project, which focuses on the economic and technical foundations of progress in computing, said in a recent MIT blog post.
One of the key takeaways from the new study is that it is only economically viable for businesses to use AI to replace 23 percent of wages that are currently paid to workers. Researchers came up with that figure by determining whether AI models could be built to complete various tasks and then estimating the cost of automating those tasks, compared with the cost of paying humans.
One example referenced in the study explored whether computer vision could replace a human when scanning ingredients at a bakery to ensure they are not spoiled.
"[The research] finds that even an AI system that is 'only' as good as a human would often be prohibitively expensive to adopt, compared with current labor costs in the U.S. The conclusion is striking: A much smaller share of the labor market is at risk of automation than direct estimations based on exposure would suggest," Antonin Bergeaud, associate professor of economics at HEC Paris, told the MIT blog.
Given those prohibitive costs, it would make sense only for some of the largest companies to turn to automation for certain tasks. As of now, a company with 5,000 employees, which is larger than 99.9 percent of U.S. firms, could automate only about one-10th of tasks in a cost-effective way, according to the study.
These findings don't negate warnings from world leaders about the scope of AI, instead offering some insight into the possible rollout. In a recent blog post, for example, International Monetary Fund chief Kristalina Georgieva states that close to 40 percent of employment across the globe has AI exposure, with as many as 60 percent of jobs in "advanced economies" at risk of some impact from AI. She concludes that AI will likely worsen existing inequalities and calls on policymakers to act quickly to mitigate the effects.
According to MIT researchers, there could initially be "a significant shock" to the labor market as certain tasks become more affordable and attractive to automate. After that shock, however, the effects of AI automation on jobs is expected to be more incremental.
There are innovations that could speed up the automation process. AI-as-a-service platforms could solve the problem of scale. And if costs associated with building out, implementing, and maintaining AI decline, that could also accelerate implementation.
So for business owners eager to innovate by using AI, researchers warn that "AI job displacement will be substantial, but also gradual," which offers time for experts and policymakers to lessen the coming blow.
Photo Credit: Getty Images.