Should copywriters, journalists, and other white-collar workers be worried?
It seems like every couple of years, a new apocalyptic report or book, or unsettling video of a humanoid robot that seems inches away from vaulting the uncanny valley comes out and rattles the American psyche. Each time, a little feeling of dread builds while we wonder whether the technology that will render our jobs obsolete has finally emerged. Usually, the panic passes after a couple of days of hand wringing and a smattering of thinkpieces, but it’s been a year and a half since ChatGPT was released and the concern has only grown.
In November 2022, when the remarkably sophisticated artificial intelligence chatbot was publicly launched, it became an overnight sensation. Within five days, the tool had reached 1 million users. (To get a sense of just how fast that is, it took Instagram about 75 days to reach that many people and Spotify about twice that amount of time, economist Jadrian Wooten, Ph.D., noted in his newsletter.) Users were astonished by just how smart it appeared to be. They used it to write jokes and poems, explain quantum mechanics, and complete college-level essays, all in a matter of seconds.
Its potential is hard to measure, but already, ChatGPT and tools like it are being used by white-collar workers to perform tasks previously done by highly skilled humans. Doctors have used it to draft letters to insurance companies, and academics to write footnotes. A judge in Colombia admitted to using the chatbot to assist in a ruling, while BuzzFeed’s embracing it to help create quizzes and other content. It’s becoming clear that AI really does have the power to reshape our labor force. The question is, to what extent?
Will AI take over our jobs?
Opinions vary widely. The most pessimistic believe it could spell the “beginning of the end of all white-collar knowledge work, and a precursor to mass unemployment,” New York Times tech columnist Kevin Roose wrote of the fear surrounding generative AI in 2022. Others have offered less spicy takes. Erik Brynjolfsson, Ph.D., a highly influential Stanford economist, told Bloomberg he thinks the chatbot will become something like a “calculator for writing” that will free us from having to perform rote tasks like writing footnotes but can’t replace “thinking and writing.”
What this technology may mean for jobs is hard to determine, says Ajay Agrawal, Ph.D., a University of Toronto professor studying the economics of machine learning and a co-author of Power and Prediction: The Disruptive Economics of Artificial Intelligence. Like Dr. Brynjolfsoon, he thinks AI-powered chatbots will come to “augment” many professional writing tasks like drafting legal briefs or press releases, which are now carried out by copywriters, paralegals, and, yes, even journalists.
Consider a journalist (ahem) reporting on the potential of AI: With astounding speed, ChatGPT can whip up a grammatically immaculate piece sourced from material around the web. But it can’t interview experts or correct itself, and it has no way of spotting a truly novel or fascinating concept as a human writer could. Just look at CNET’s controversial robot writing experiment, which resulted in several incidents of plagiarism and some substantial corrections. A chatbot still needs human supervision, but it may make a journalist faster and more efficient.
ChatGPT is not only a competent scribe, but also a proficient computer programmer. It can write lines of script and explain how it works almost instantly, but it has its limitations. It can’t fix certain bugs or wholly design software that considers the complicated demands of a particular business, one data scientist points out. Even still, it’s already being used by developers as a shortcut.
It’s “unquestionably quicker to tell ChatGPT what you need and then check the code than it is to manually write everything from scratch,” writes one engineer, who also urged others to treat the AI “as a personal assistant.”
That’s why, at present, Dr. Agrawal and other notable economists think ChatGPT won’t result in the wholesale destruction of certain roles just yet. But if it makes workers more productive, it may mean that companies find they need fewer employees, which could ultimately “result in people losing jobs,” Dr. Agrawal told us last year.
Last year, Pew took a crack at measuring the threat AI could pose to different sectors. It found that 19 percent of American workers have jobs that are highly exposed to the technology — gigs like budget analysts, data entry specialists, tax preparers, technical writers, and web developers. It also found that more educated people (folks with a bachelor’s degree or higher) were twice as likely to work in one of these jobs. The jobs least at risk, as you may guess, are roles like barber, firefighter, or childcare worker, which simply can’t be replaced or augmented in any significant way by a chatbot.
Could AI create jobs?
It may create jobs, too. For instance, by using the customer-service chatbot Ada, Zoom was able to quickly scale up during the pandemic, bringing on a ton of new business and creating thousands of new jobs as companies turned to the app to facilitate remote work. This is what’s referred to in economics as the scale effect, Dr. Wooten, a Pennsylvania State University professor, wrote.
There’s also Uber to consider: The company harnessed navigational AI platforms like Google Maps and Waze to “upskill” millions of amateur drivers, Dr. Agrawal said. Suddenly, anyone could determine the fastest route from midtown Manhattan to Hoboken during rush hour just by looking at their phone; they didn’t need years of experience driving in New York City. On one hand, Uber and Lyft have obviously created millions of part-time jobs, but they’ve also upended the taxi industry and left many full-time cab drivers a lot worse off than they were before the apps.
That phenomenon has worried a set of economists, who fear generative AI could do something similar but on a much larger scale. It’s possible tools like ChatGPT, Microsoft’s Bing AI, or even Google’s imperfect Bard could shake up the composition of jobs even in fields previously considered immune to automation. It may create new opportunities for the college-educated and raise living standards, or it may limit those secure, middle-class gigs — pushing more people into lower-earning roles. If that were the case, it would only further widen the already massive wealth gap here in the U.S., Lawrence Katz, Ph.D., a Harvard labor economist, told the Guardian.
Of course, no one has a crystal ball, and how AI will change the employment landscape is really anyone’s guess. But what we do know is that the startling capacity of this tool has finally woken the public up to AI’s tremendous potential and that companies are now scrambling to figure out if it’s a threat or something that can give them a competitive edge, Dr. Agrawal said. In the weeks after ChatGPT launch, he told us his phone had been “ringing off the hook,” with CEOs from all types of industries trying to read the tea leaves.
“They don’t want to get left behind.”
