After analyzing 5 million freelance jobs, it was found that these jobs are most likely to be replaced by AI

8 min readJun 24, 2024

This article explores which jobs are more likely to be replaced by AI by analyzing 5 million freelance jobs on the Upwork platform from November 1, 2022 to February 14, 2024.

Key findings

Job categories with the largest decreases:

Writing jobs fell by 33%.
Translation work decreased by 19%.
Customer service jobs fell 16%.

Job categories least affected:

Video editing/production jobs increased by 39%.
Graphic design jobs increased 8%.
Web design jobs increased by 10%.
Software development jobs also increased, with back-end development up 6% and front-end/web development up 4%.

Job categories with the biggest declines in hourly wages:

Hourly wages for translation jobs have fallen by more than 20%.
Hourly wages also fell for video editing/production and market research jobs.
Graphic design and web design jobs have not only increased in number, but also in hourly wages.

Increasing demand for AI skills:

There is an increase in job demand for AI-related content generation, AI agent development, OpenAI/ChatGPT API integration, and AI application development.
The demand for jobs related to chatbot development has surged by 2,000%.

There’s no doubt that artificial intelligence (AI) will have an impact on jobs. But which jobs are more likely to be replaced by AI? Which jobs are already being replaced by AI? To find out, I decided to look at hard data : actual freelance job data on Upwork from November 1, 2022 (a month before ChatGPT was released) to February 14, 2024 to see which jobs were most negatively impacted.

Why choose to analyze freelance jobs instead of actual jobs at big companies like Microsoft and Amazon? There are two reasons. First, if certain jobs will be affected, we may see it first in the freelance market, as big companies are slower to adopt AI tools. Second, judging by Upwork’s latest earnings report , layoffs in the tech industry haven’t had much of an impact on the Upwork freelance marketplace. In the most recent quarter, their revenue grew 14% year-over-year. This means that the decline in job postings isn’t affected by a lot of other factors.

Note : All of this data is provided by Revealera.com , a data provider that provides company hiring trends data. Upwork’s job posting data is also all publicly available in the form of an RSS feed, so anyone can easily analyze the data.

Here’s what you’ll learn from my research:

Which job categories have seen the biggest declines?
Which job categories are least affected?
Which job categories saw the biggest declines in hourly wages?
Which AI skills have seen the largest increase in job postings?

Which job categories have seen the biggest declines?

I selected the 12 most popular job categories on Upwork and analyzed the 84-day moving average of these categories over a period of approximately 3 months. Surprisingly, most job categories have actually seen an increase in job count since ChatGPT was released , but 3 categories have seen a significant decrease in job count.

The three categories that saw the biggest declines in job numbers were writing, translation, and customer service jobs. The number of writing jobs fell by 33%, translation jobs fell by 19%, and customer service jobs fell by 16%.

Which job categories are least affected?

Aside from the few job categories mentioned above, most other job categories were not negatively impacted — in fact, the number of jobs increased.

Since the launch of ChatGPT, the number of video editing/production jobs has increased by 39%, graphic design jobs have increased by 8%, and web design jobs have increased by 10%. Software development jobs have also increased, with back-end development jobs increasing by 6% and front-end/web development jobs increasing by 4%.

On the surface, some of these changes are not surprising. I did expect writing jobs to decrease, as this is probably the most popular use case for ChatGPT, which is reflected in the 33% decrease in writing jobs. Likewise, it is not surprising to see customer service jobs decrease by 16%, as many companies are creating chatbots that can replace many customer service agents.

On the other hand, I was surprised to see an increase in the number of jobs in graphic design, video editing/production, and even software development, considering we heard a lot of stories about people using ChatGPT to generate code, illustrations, and even full videos.

I think there are a few ways to explain this data. One possibility is that these generative AI tools are good enough to replace many writing tasks, whether it’s writing articles or social media posts. But for other tasks, such as video and image generation, they are not perfect yet. Personally, I need many prompts before I am satisfied with the images that DALL-E creates for my blog posts, and that’s just because I’m not too picky.

So, if you’re creating images that require a lot of detail, these AI tools may not be good enough yet. Similarly, for video, we’ve seen some tools that seem to be able to make many complex video scenes, but can they really create a full trailer or customer tutorial video? Not yet, although this may change in a few months .

Another explanation is that it takes time for users to learn how to use these tools effectively and understand their capabilities and limitations. If we look at the graph of the decline in writing jobs, we see that it started to decline soon after ChatGPT was released, but the rate of decline did not accelerate until August (almost ten months after ChatGPT was released).

Which job categories saw the biggest declines in hourly wages?

Is there a correlation between the decline in the number of jobs and the decline in pay? Unfortunately, there is no public way to know exactly how much freelancers are actually paid, so for this research I had to extract the estimated hourly pay range specified in the job postings , which would be a close estimate of what freelancers are actually earning.

As shown in the chart, translation jobs were the hardest hit, with hourly wages dropping by more than 20%, followed by video editing/production and market research. Graphic design and web design jobs were the most resilient. Not only did they increase in number, but hourly wages also rose slightly.

Overall, I think it’s hard to draw firm conclusions from this data. Many of the salary changes may just be normal seasonal fluctuations. However, one clear conclusion is that graphic design and web design jobs are still in demand and have not yet been replaced by AI tools.

Again, I think this is because tools like DALL-E and MidJourney require some knowledge and creativity. This is evident in some job postings that specifically ask for AI image creation experts. Unlike writing articles, creating visually stunning images using AI tools requires understanding how to construct good prompts, how to adjust the prompts, and a lot of trial and error.

Which AI skills have seen the largest increase in job postings?

Okay, so we know which jobs are declining in demand, but which ones are actually increasing in demand due to AI?

My first reaction is that there should be an increase in demand for data annotation jobs to help train and fine-tune large language models (LLMs), and freelancing platforms are undoubtedly the best choice for finding people to annotate data. While I don’t expect most companies to hire many machine learning experts on Upwork, the number of machine learning jobs should at least increase, right?

It turns out that both of these assumptions were completely wrong . While there was an initial increase in data annotation jobs, the number of data annotation jobs has remained flat over the past 10 months. And the number of machine learning jobs has actually decreased since ChatGPT was released.

Perhaps companies are hiring more machine learning engineers outside of Upwork. After all, who would outsource important, specialized machine learning work to a freelancer? Yet, even when we look at the world’s largest companies, there has not been a significant increase in demand for machine learning.

On the other hand, jobs for generating AI content, developing AI agents, integrating the OpenAI/ChatGPT API, and developing AI applications are becoming increasingly popular. But what is the hottest use case? Chatbots. Since the release of ChatGPT and the OpenAI API, the number of jobs related to developing chatbots has surged by 2,000%. If there is a killer use case for AI today, it is developing chatbots.

I think one takeaway from this is that most companies aren’t doing the exciting work of developing their own Large Language Models (LLMs) or fine-tuning them with lots of training data. Instead, they’re more about integrating OpenAI’s API into existing products and developing chatbots to replace customer service agents.

In conclusion

Since the launch of ChatGPT, there has been a noticeable decrease in writing, customer service, and translation jobs on Upwork.
Web design, graphic design, software development and video production jobs were the most resilient, with demand not decreasing (and even increasing).
It’s unclear whether this is because AI tools aren’t good enough yet, or because the public hasn’t learned how to use them effectively.
Judging by the lack of growth in data annotation or machine learning work, most companies are not currently focused on training or fine-tuning their own large language models.
Currently, the most popular use case for AI is undoubtedly the development of chatbots.