Why humans shouldn't fear being replaced by AI automation

With businesses now deploying artificial intelligence tools to do everything from translating speeches to building language learning models for large legal firms, many employees are starting to worry that they may soon be out of a job.

However, Noah Smith, author of the popular Noahpinion website, argued in an article that people shouldn’t worry about their jobs being automated just yet.

The common perception of automation is that it will drive humans out of the workforce—one day you are doing some valuable work, tomorrow you are on welfare,” Smith wrote. However, this is not the case! We have been deploying automation technology for centuries, and as of 2023, almost everyone who wants to work will still have a job."

What does automation mean?

In his article, Smith looked at several studies on job automation over the years by researchers from Citibank to PricewaterhouseCoopers.

Smith points out that in these studies, the term "automation" is never clearly defined, but together they all illustrate varying degrees of "substitution" assumptions. In some cases, like "You'll get new tools that let you automate boring parts of your job, allowing you to be promoted to a more valuable position," there will even be new benefits.

This means it’s difficult to draw sweeping conclusions about what automation will mean for any particular individual.

Another problem Smith points out is that these studies do not address how the labor market as a whole will change. He believes that "if one job is destroyed by automation and two other higher-paying jobs are created, the worker is the clear winner." However, research on this topic seems to focus only on automation, which may indicate that workers are The loser in this situation, even if that's not the case.

Assessing “replacement” is often subjective

Smith also pointed out that previous studies have used too much subjectivity when assessing the risk of a job being replaced.

In a study by Frey and Osborne (2013), the researchers noted that they "subjectively hand-labeled" jobs from a database developed by the U.S. Department of Labor if they were "automatable" , then score 1 point, if not, score 0 points.

The researchers also noted that they focused on only a small subset of the jobs in the database "for which we have high confidence in their computerized labels" to further reduce the risk of "subjective bias affecting our analyses."

The good news is that research methods have improved since then, Smith said.

A Goldman Sachs study published earlier this month assessed the impact of artificial intelligence on automation by looking at jobs as a sum of tasks described in government databases, rather than as a monolithic entity.

Goldman Sachs researchers also recognized that when only some tasks are automated, "automation often serves a complementary role rather than displacing workers," Smith noted.

Additionally, the study supports the idea that automation doesn’t always mean layoffs, noting that “technology can replace some tasks, but it can also make us more efficient at performing others and create New tasks and new jobs.”

However, when Forbes reported this study, the headline was "Goldman Sachs predicts artificial intelligence will eliminate or reduce 300 million jobs."

Smith writes, "Many people are so used to the 'robots taking our jobs' narrative that they report every outcome they see through a distorted lens.'"

The above is the detailed content of Why humans shouldn't fear being replaced by AI automation. For more information, please follow other related articles on the PHP Chinese website!