GenAI is already driving down demand for freelancers

The demand for freelancers working in writing, and coding related jobs has dropped by 21 percent since the launch of ChatGPT in November 2022, according to new research by Imperial College Business School, Harvard Business School and the German Institute for Economic Research. This research paper, Who is AI Replacing? The Impact of Generative AI on Online Freelancing Platforms, looks at the impact of GenAI technologies on the demand for creative professionals who work on an hourly basis for different clients.

The researchers reviewed data on a global online freelancing job platform, of almost two million job postings on the site, spread across 61 different countries, for a two-year period between July 2021 – July 2023.

They categorised these jobs roles into three sectors – automation-prone, manual work and image generation. The researchers reviewed the frequency of job postings throughout the two-year period to understand the demand for freelancers working for different organisations, and whether this was influenced by the launch of Chat GPT.

The study found that since the introduction of ChatGPT in November 2022, the number of job postings related with writing, software, app and web development, and engineering, otherwise known as automation-prone jobs, had fallen by 21 percent compared to the period before ChatGPT was launched.

However, freelance job postings focused around data entry, video and audio editing, and social media post production were slightly more less affected by ChatGPT, dropping by 13 percent. Jobs focused on image-generation such as graphic design and 3D modelling saw a 17 percent drop in demand on the online job platform after image-generating AI tools.

The researchers backed up these findings by reviewing patterns from Google search volume indexes, which confirmed that sectors where workers had a high awareness on generative AI and how to utilise it i.e. the tech industry, exhibited a more significant decline in demand from employers.

It still remains to be seen whether organisations are happy with the quality of work that AI provides

Dr Xinrong Zhu, co-author and Assistant Professor of Marketing at Imperial College Business School, said: “Despite being available on the market for just over a year, ChatGPT has already had a huge impact on the workplace. Though many organisations may be shifting from freelancers to generative AI, it still remains to be seen whether organisations are happy with the quality of work that AI provides in comparison to freelancers, and whether this trend will continue.

“Although the findings of our research suggest that the jobs market looks gloomy, whenever technology sweeps professions aside, new ones will emerge. For freelancers, this means that those people adapt their skillsets to the changing landscape, will continue to secure work in the future,” she said.

The researchers say that it remains unclear how generative AI technology will affect the digital labour market in the long run. The widespread adoption of generative AI by employers as a replacement for human workers may exacerbate an existing decline in demand for labour, worsening workers’ welfare. Meanwhile, the research suggests that generative AI can also enhance workers’ productivity and could consequently improve people’s earnings.

Looking ahead, the researchers say it is important to emphasise the ongoing need for roles that demand a more human touch, such as projects that call for a greater degree of creativity, intricate problem-solving and nuanced understanding. According to the researchers, the future workplace will require a multitude of different skills and the human intelligence to provide them.

This research was conducted by Dr Xinrong Zhu, Assistant Professor in Marketing at Imperial College Business School, alongside Dr Ozge Demirci, a Postdoctoral Fellow from Harvard Business School and incoming Imperial Assistant Professor, and Jonas Hannane, a PhD student from the German Institute for Economic Research. The study is available to download from the SSRN website.