- As AI tools revolutionize business, workers are worried they're at risk of losing their jobs.
- Yet Noah Smith, who writes the Noahpinion newsletter, contends there are nuances amid the expected impact.
With businesses now deploying AI tools to do everything from translating speech to building language learning models for big law, many workers are worried they might soon be out of a job.
Yet Noah Smith, the writer behind the popular Noahpinion economics newsletter, contended in a post on Monday that people shouldn't worry about losing their jobs to automation just yet.
"The "folk model" of automation is that it throws humans out of work — today you had a job performing some sort of valuable work, and tomorrow you're on the welfare rolls," Smith wrote.
"We've been deploying automation technology for centuries, and as of 2023, pretty much every human who wants a job has a job," Smith wrote, "but there's basically no way to get people to believe that this next wave of automation will be the one that finally sends humans into obsolescence."
What exactly does automation mean?
In his post, Smith examined several studies on job automation over the years from researchers at firms ranging from Citibank to PriceWaterHouseCoopers.
Smith points out that the term "automation" in these studies is never clearly defined and together they illustrate hypotheticals that poses different levels of "replacement." Some scenarios, such as "You're going to get new tools that let you automate the boring part of your job, move up to a more responsible job title" even present newfound benefits.
That means it's hard to draw sweeping conclusions about what automation means for any particular individual.
The other issue Smith pointed out is that these studies don't touch on how the labor market will change overall. "If one job is destroyed by automation and two more are created for higher wages, workers obviously won out," he suggested. Yet studies on the topic only seem to focus on automation, which can suggest that workers are the losers in the situation, even if that's not really the case, Smith wrote.
Assessing "replacement" is often subjective
Smith also pointed the subjectivity used in older studies for assessing a job's risk of replacement.
In one study, Frey and Osborne (2013), researchers noted that they "subjectively hand-labelled" jobs from a database developed by the US Department of Labor with a score of 1 if they were "automatable" and 0 if they were not.
The researchers also noted that they only focused on a small percentage of jobs in the database "whose computerisation label we are highly confident about" in order to further reduce the risk of "subjective bias affecting our analysis."
The good thing, Smith said, was that studies have improved their methodology since then.
A study by Goldman Sachs published earlier this month assessed AI’s impact on automation by viewing jobs as a sum of tasks described in a government database as opposed to a holistic entity.
Smith noted that Goldman's researchers also recognized that "automation often ends up complementing a worker's effort instead of substituting for it" when only some tasks are automated.
Added to that, the study supported the point that automation doesn't always spell layoffs, noting that "technology can replace some tasks, but it can also make us more productive performing other tasks, and create new tasks — and new jobs" which supports the point that automation doesn't always spell layoffs.
Yet Forbes reported on the study with the headline "Goldman Sachs Predicts 300 Million Jobs Will Be Lost Or Degraded By Artificial Intelligence."
"A lot of people are so used to the "robots take our jobs" narrative that they report every result they see through that warped and distorted lens," Smith wrote.
Smith did not immediately respond to Insider's request for a comment.