So, Yeah, AI Is Already Taking Our Jobs
A new study sheds light—or rather shadow—on the job question
Automation wasn’t invented in the 19th-century factories of the Industrial Revolution, nor during the Italian Renaissance of the 16th century. The idea of machines imitating human actions goes back much further. We have to travel two thousand years into the past, to the 3rd century BCE, during the Hellenistic period—a time long before even the steam engine, which some believe was first developed in a rudimentary form by the Romans.
In the sunlit archipelagos that populate the Aegean Sea southeast of what’s Turkey today, you’d find the influential maritime power of Rhodes. And somewhere between Rhodes and the Library of Alexandria (depending on when you arrived), you might encounter the renowned poet and scholar Apollonius. While he didn’t invent the concept of automation, he’s credited with one of the earliest—and most influential—stories of machine’s power over man.
I’m referring to the myth of Talos, the bronze automaton who guarded Crete, tasked by Hephaestus to fend off invaders, pirates, and any sailor with questionable intentions. It didn’t last long—or so Jason and his Argonauts might recount today. They confronted and ultimately defeated the giant sentinel, but only with the help of the sorceress Medea, who advised Jason to “fight Talos with your wits rather than your courage.”
The 1963 film adaptation of Argonautica by Apollonius of Rhodes left a lasting impression on me—that one line especially. It’s the only line I can still recall, and I won’t be watching it again. That movie filled my nights with horrors, much as I imagine Talos scared the brave Argonauts themselves. Perhaps it was the unsettling, uncanny movements of the towering colossus as it chased the poor sailors, or perhaps the chilling realization that if humanity can imagine something, it can bring it to life.
Whatever the case, I have to admit my old nightmares are nothing compared to those inspired by generative AI today, two thousand years later, which haunt the unwitting workers of the 21st century, living perpetually on the edge of replacement. Like the Argonauts, they are equipped with both ingenuity and courage, yet they lack a Medea by their side to cast the right spell.
However, when I consider the effect of AI on the modern workforce, I always return to a conclusion that feels both grounded and, surprisingly, hopeful—one that counters the nightmarish visions with a sense of possibility.
You using AI will replace yourself not using it.
In the long run, the novelty of AI turns into custom, hype fades away, and the rightful resistance of the defendants of the status quo gives in to the benefits of an emotionally discharged innovation: You will use AI because eventually, you won’t be mad at it.
As I said, I’m optimistic.
In the short term, however, not so much. Another, more popular slogan shapes my predictions:
AI won’t take your job, a person using AI will.
A lot of people will lose their jobs because demand for their skill set can’t grow as fast as AI-induced productivity. Contrary to what the myth of Jason and the Argonauts may suggest, humans don’t usually come out victorious in battles against automation.
Unfortunately, my short-term concerns are true already. A recent Harvard Business Review study explores the pressing question: What impact is generative AI currently having on the labor market? The findings are bleak.
Researchers took data from an (undisclosed) popular freelance platform (think Upwork or Fiverr) and measured the number of job postings before and after OpenAI released ChatGPT in November 2022. They compared two categories of jobs, automation-prone (e.g., writing; software, app, and web development) and manual-intensive (e.g., data and office management, video services, and audio services). (They also analyzed AI image tools but I will focus on ChatGPT here for the sake of simplicity. The results are similarly gloomy.)
The findings are shown in the graph below. Look at the downward trend (that admittedly was already present before ChatGPT, possibly due to the existence of GPT-3-like language models and generative AI writing startups like Jasper and Copy.ai):
Here’s the quantitative data from the study:
After the introduction of ChatGPT, there was a 21% decrease in the weekly number of posts in automation-prone jobs compared to manual-intensive jobs. Writing jobs were affected the most (30.37% decrease), followed by software, app, and web development (20.62%) and engineering (10.42%). . . . Additionally, we noticed that over time, there were no signs of demand rebounding.
30% fewer “writing job” postings after ChatGPT compared to the control group (manual-intensive) which also decreased between December 2022 and July 2023. Not good. The study concluded that automation driven by generative AI tools has been faster and more drastic even than robotic automation:
. . . for example, researchers found that a 20-percentage-point increase in robot adoption in French manufacturing led to only a 3.2% decline in industry employment.
The silver lining? If you know how to use generative AI tools, the future looks brighter:
The number of job posts seeking “ChatGPT” skills has grown steadily . . . suggest[ing] that the ability to integrate AI tools into work is becoming increasingly valued, and workers are likely updating their skill sets to include gen AI capabilities.
So: ChatGPT has made a noticeable impact on the online office-type workforce, posing a challenge for writers, coders, designers, and related fields. Even for those who adapt to using generative AI, the new skills don’t offset the decline in demand.
The data are compelling but I must offer a word of caution: further research is needed. Let’s not be more pessimistic than strictly needed.
Until then, a few questions arise that I bet you’d like to see answered: what do the findings imply more broadly, besides fewer job postings on Upwork and Fiverr? What should writers, designers, and coders do to fight back against automation? Is the trend going to continue in the mid and long term? How does this translate to people employed by others (in contrast to self-employed)? Is there any hope for creativity?
I will respond with five brief points, which, as a writer myself, will define my approach and my attitude toward the present and beyond: