Surprise, surprise: the data shows the impact of ChatGPT on the creative professions

Summary

Recent academic studies reveal that ChatGPT has negatively impacted creative professions such as graphic design and copywriting, with a decrease in job opportunities and earnings.

Abstract

Researchers from Washington University, NYU, and Harvard have conducted studies on the impact of ChatGPT on employment, particularly in creative fields. The data, collected from freelancing platforms like Upwork, indicates a significant reduction in job postings and a steeper decline in earnings for graphic designers and copywriters since the introduction of ChatGPT. The study suggests that generative AI not only replaces human jobs but also devalues the remaining work. High-earning, skilled workers are not immune to these effects, as the demand for knowledge workers across the board has decreased. These findings are crucial for understanding the short-term effects of AI on the job market and could influence future regulations on AI tools.

Opinions

The introduction of ChatGPT has led to a direct substitution of human labor in creative professions.
Generative AI like ChatGPT is seen as diminishing the value of human-generated creative work.
The negative impact on employment and income is not mitigated by higher skill levels or a history of higher earnings.
The study underscores the role of academics in providing concrete data to support observations about AI's impact on the workforce.
The widespread adoption of generative algorithms post-ChatGPT's launch has created a clear timeline for measuring societal effects.
There is an expectation that more research will follow, contributing to predictions about AI's influence on various industries and labor markets.
The quantifiable effects of generative AI could lead to future regulations governing the use of such technology.

Surprise, surprise: the data shows the impact of ChatGPT on the creative professions

IMAGE: Mohamed Hassan — Pixabay

The study’s conclusions are hardly surprising, but which someone had to provide concrete data for, and that’s what we academics are for: the introduction of the ChatGPT generative algorithm a year ago has had significant negative effects on creative professions such as graphic designers and copywriters.

Using data from copywriters and graphic designers on the freelancing platform Upwork, the researchers found that these professionals not only experienced a significant drop in the number of jobs they could bid for, but even steeper drops in earnings, suggesting not only that generative AI is taking away their jobs, but that it tends to devalue the dwindling volume of work they still do.

The case of copywriters and graphic designers is particularly interesting because it illustrates the phenomenon of direct substitution: the work that someone used to request from one of these professionals, even considering the possible iterations until a satisfactory result was obtained, can now be carried out directly by the generative algorithm. The results are significant and amount to around 2% in the number of jobs obtained monthly, and 5.2% in monthly income from the time ChatGPT was released.

Moreover, workers who previously typically earned the highest incomes and who completed a greater number of jobs do not result in a lower likelihood of seeing their employment and income decline, but even worse outcomes, implying that having more skills is no protection against job or income loss. These results suggest that, in the short term, generative AI reduces the overall demand for knowledge workers of all types, and may also have the potential to reduce gaps between workers.

As time goes on, we will undoubtedly see more academic work trying to quantify the effect of the emergence of generative algorithms. Their widespread adoption following the launch of ChatGPT over the last 12 monthsprovides a discontinuity in time that makes it easy to quantify the effects on individual practitioners. Moving from the realm of untested hypotheses to concrete and specific quantification and proving that there are indeed significant effects on society is a first step towards making predictions about different industries and labor markets, and may well affect the regulation of the use of such tools in the future.

(En español, aquí)