The Economic Impact of Artificial Intelligence on the Labour Market and Economic Growth

The rise of generative artificial intelligence (AI) has caused quite a stir in recent years. Especially in the last months, with the emergence of generative AI models like ChatGPT, DALL-E or Midjourney, the question has been raised: How many jobs are going to be replaced by AI? And what impact does it have on our economy? Or ultimately: Will it be a good or a bad thing?

In this article, we review a few key takeaways on the economic impact of AI as reported by analysts from sources like Goldman Sachs [1], pwc [2] as well as scientific research [e.g. 3] to understand the scale of impact of the latest advances in the field of artificial intelligence and especially generative AI.

A Revolution?

Generative AI tools such as ChatGPT, DALL-E, and Midjourney have seen a rise in popularity in recent years due to their ability to automatically generate content that often can hardly be distinguished from human-created content.

Text-based generative AI tools like ChatGPT, for example, can be used by companies in the marketing industry to automate tasks such as writing interesting and engaging content for social media posts or creating customer support scripts. In the legal field, it can also be used to generate summary reports of larger text passages quickly. Product manufacturers can swiftly create product descriptions or catalog entries.

Visual generative AI tools like DALL-E or Midjourney can generate photorealistic images and videos from text descriptions, allowing users to quickly create content such as graphic design, advertising posters, logos, or short video clips.

In the past months, we could see many businesses and startups diving into building applications on top of these generative AI tools for specific use cases. But this "AI buzz" that can be felt in recent months is nothing that came suddenly. Over the past decade, the advent of deep learning approaches was possible due to a fast-changing raw computing power [4] that made the technical concepts like deep learning that have been around for decades finally practical.

The new wave of such generative AI tools has such a disruptive character because of two key advances:

1. The ability to generate new content that is indistinguishable from human-created output and
2. The ability to break down communication barriers between humans and machines through the use of natural language interfaces

These two aspects provide an economic opportunity, as companies would have access to a more efficient means of creating customized content than creating that through manual labor.

The Impact on Work as we know it

Goldman Sachs economists [1] estimate that as many as 300 million full-time jobs worldwide could be automated in some way by the newest wave of artificial intelligence that has spawned platforms like ChatGPT.

According to their report, 18% of work globally could be fully automated. The impact will be more significant in advanced economies compared to emerging markets. One reason is that white-collar workers are perceived to be more vulnerable to automation than manual laborers. The economists predict that administrative workers and lawyers will be most affected. In contrast, it is expected to have a "little effect" on physically demanding or outdoor occupations, such as construction and repair work.

The bank estimates that about two-thirds of current jobs in the United States and Europe have some level of exposure to AI automation. AI could potentially completely take over up to a quarter of all work. This could be a significant disruption to the global labor market.

Job Losses or Augmentation?

So is AI replacing you in doing your job? Well, partially. "Although the impact of AI on the labor market is likely to be significant, most jobs and industries are only partially exposed to automation and are thus more likely to be complemented rather than substituted by AI" the Goldman Sachs economist state in their report.

Therefore, we have to distinguish between generative AI tools working as a complement for existing tasks for some occupations and completely replacing manual labor by automating the respective tasks entirely. When we look, for example, into office and administrative support jobs, Goldman Sachs estimates that about 35% of the tasks in such jobs can be automated entirely. In comparison, about 55% of tasks for administrative workers can be complemented by using AI tools. The remaining 10% are likely not exposed at all to automation.

This lets us group occupational fields into three different groups:

• Largely AI-complemented fields. Work activities in industries like computer science, software engineering, educational instructions, community and social services, businesses and financial operations will likely be to > 90% be complemented by AI, yet almost not all fully automated. Software engineers, for example, can use AI tools to enhance their workflow greatly. ChatGPT and GitHub Co-pilot can already generate snippets of code. Yet having complete complex systems of software generated by an AI without any human "stitching parts together", doing quality assessments and looking into the most demanding tasks are not practical in the near to mid-term future. Thus, the work of a software engineer will not go away. It will be just different.
• Largely automated fields: Work activities in legal or administrative support fields will have a large percentage of workflows (32-40%) fully automated. The remaining 50-60% in such fields can additionally be complemented by the use of AI tools by workers. Typical administrative tasks like planning meetings can already today be fully automated (in that case, with tools like Calendly). Thus, in that field, we have a situation where AI entirely removes for a percentage of their work the need of a human.
• Largely unimpacted fields: Industries with a lot of physical labor, for example, in construction, building, installation and repair industries, will only see a small percentage of up to 30% of their work activities being complemented by AI.

Chances for Productivity and Growth

Using generative AI for automation could significantly increase global labor productivity, leading to a potential rise in the global GDP. According to two Goldman Sachs, this will go through two main channels:

First, most workers work in jobs that AI will partially automate. They will probably use some of the time they freed up from automation to do other productive tasks that increase productivity. Here we can see how highly skilled and experienced workers can increase their output by automating / "letting AI" do the repetitive tasks while they focus on applying their specific knowledge to fields where AI cannot help yet.

Second, the economists predict those who lose their jobs due to AI automation will eventually find new employment. This will, in turn, increase overall productivity. These new jobs may arise either from the increased adoption of AI technology or the higher demand for labor resulting from the improved productivity of non-displaced workers.

Tieing it all together

What can we learn from this?

"The combination of significant labor cost savings, new job creation, and a productivity boost for non-displaced workers raises the possibility of a labor productivity boom like those that followed the emergence of earlier general-purpose technologies like the electric motor and personal computer." [1]

Many industries will see a significant impact on their jobs. This happens either by complementing work with AI tools or by completely automating tasks. The work as we know it will change. The impact of that change will increase our overall productivity. Yet, we have to remind us that these are statements are predictions, not facts. It is important to stay careful and aware of the potential impacts that come with the adoption of AI technology - and in the end, make the best out of it for all.

References

[1] Hatzius, Jan, Briggs, Joseph, Kodnani, Devesh and Pierdomenico, Giovanni "The Potentially Large Effects of Artificial Intelligence on Economic Growth", Goldman Sachs Economics Research, March 2023

[2] pwc, "Sizing the prize, PwC’s Global Artificial Intelligence Study: Exploiting the AI Revolution",  2017

[3] Aghion, Philippe, Benjamin F. Jones, and Charles I. Jones. "Artificial intelligence and economic growth." The economics of artificial intelligence: An agenda. University of Chicago Press, 2018. 237-282.

[4] Sevilla, James et al., “Compute Trends Across Three Eras of Machine Learning”, 2022 International Joint Conference on Neural Networks, 9 March 2022.

[]The Potentially Large Effects of Artificial Intelligence on Economic Growth

‍