Ever since ChatGPT hit the scene a couple years ago, there’s been a nagging sense of dread for many: what will this mean for jobs? Well, new research from Imperial College London finds a shift already underway.

Between July 2021 and July 2023, the report found freelance job postings for writing and coding decreased by about 20%. There was also a slowdown in freelance jobs for visual art.

And it’s happening more quickly than past technological disruptions, Ozge Demirci, one of the coauthors of that report and a business professor at Imperial College London, told Marketplace’s Meghan McCarty Carino. The following is an edited transcript of their conversation:

Ozge Demirci: Compared to past automation technologies, I think generative AI has the unique potential to impact all the sectors. For instance, comparing generative AI’s impact to robotic automation, in those cases researchers found that 20 percentage point increase in robot adoption in manufacturing led to only around 3% decline in industry employment. This is not one-to-one comparison, but even looking at the magnitudes, effects, we can say that generative AI is much more far reaching and is coming for almost all jobs.

Meghan McCarty Carino: And the speed of these kinds of changes tends to be important when it comes to things like upskilling, reskilling. When these changes happen sort of over longer time periods, there is a bit more opportunity to adjust to the labor market, right?

Demirci: Yeah, definitely. I mean, our study focuses on, we can say that we only look at the short term impacts within the eight months after ChatGPT’s introduction. And we can claim that it’s just the earlier response of the markets. We see a big decline in the demands, but we also start to see the content of the jobs are also changing. Obviously, there are also other impacts rather than just decline in the demand.

McCarty Carino: Yeah, let’s talk about some of those other impacts. You found some changes to competition and pay also in many of these jobs.

Demirci: Yeah. On the one hand, AI is definitely replacing some jobs, especially in areas that can be easily automated. Writing is a prime example. It’s one of the most affected fields in our study. But on the other hand, AI is also opening up new possibilities for workers who can adopt and incorporate these tools into their skillsets. So one thing we noticed is increased competitions for the jobs that remain; for instance, after ChatGPT’s launch, the average number of applications per job in writing and coding increased by 9%. So it means that it’s getting more and more difficult to secure work in these areas, but at the same time employers are willing to pay more for these jobs, because possibly these jobs require more complex skillsets or specific AI expertise. So the market is evolving to favor those who can upskill and take on these more advanced roles.

McCarty Carino: So what does that tell us about how workers might prepare for these disruptions?

Demirci: I think in the short term, we can say that it does look like there is a replacement, and it’s happening more than, let’s say, augmentation, especially in automation-prone areas. But when you look a bit deeper, there is evidence that augmentation is also happening. For example, jobs that explicitly require skills, like using ChatGPT, are on the rise as well. And this can imply that workers who can integrate AI tools into their skillsets will not be replaced. They are just driving in the new ways. And as these technologies evolve, they can actually free people from repetitive tasks and allow them to focus more on creativity and strategic thinking. So this kind of augmentation, it just doesn’t happen on its own. It requires a certain effort by organizations, because they need to now reskill their workforce and integrate AI thoughtfully.

McCarty Carino: We talked about what this might mean for workers. What does this mean for firms?

Demirci: For firms, I think it definitely requires a certain effort to reskill their workforce. And I think companies that invest in training and focus on combining human creativity with AI efficiency will be the ones most likely to succeed in the long run. It’s not a simple augmentation-or-replace scenario, but rather generative AI has potential to enhance human capabilities, but only if workers, and especially organizations, take steps to adapt to this new reality.

McCarty Carino: I think for a lot of us working in creative endeavors — writing, art, etc — your research, sounds pretty depressing.

Demirci: No, not at all. I mean, yes, I totally feel it. But I think from now on, the ones who will be maybe a bit at an advantage stage will be the ones who can adopt ChatGPT, or who can adopt this new generative AI skills, and make these skills part of their skill set. It can even increase the productivity of many of us, actually, if we can manage to use these technologies in an efficient and in a correct way. So yeah, at the beginning, it may look a little pessimistic, but if these tools are used in a correct way, it can be just very helpful for all of us.