The Impact of LLM on Jobs: What’s Happening in 2023

Introduction:

A great number of studies have pointed out the risks of AI to the job market. With the widespread adoption of Large Language Models (LLMs) and Generative AI tools in 2023, many people feel that job security is at risk. However, it’s essential to note that while concerns about AI’s impact on employment are prevalent, empirical studies have yet to substantiate these risks with concrete numbers of job losses. In this blog post, I will aim to bridge this gap by examining the existing research and shedding light on the evolving relationship between LLMs, Generative AI, and the job market in 2023.

The Challenge of Studying AI and Jobs

Understanding the connection between AI and employment is a complex task. Several challenges make it hard to pinpoint precise cause-and-effect relationships. Here are some key challenges:

1. Time: Effects of AI on jobs can take years to become evident, making it challenging to link current developments to immediate job market changes.
2. Data Availability: Accessing consistent, high-quality data on AI adoption and employment trends can be a challenge, especially when focusing on the most recent developments.
3. Causality: AI adoption isn’t the sole factor influencing job displacement. Economic, political, and industry-specific variables play significant roles.
4. Sector Variability: The impact of AI varies widely across different industries and job sectors. This makes it crucial to analyze specific sectors when assessing the impact on employment.
5. Indirect Effects: AI can both displace and create jobs, making it challenging to quantify its net effect on employment.
6. Global Variation: The effects of AI on employment can vary significantly from one country to another, depending on factors like language and technological adoption.

In (Mutascu, 2021), the author embarked on a quest to unveil the intricate, non-linear relationship between AI adoption and its impact on the job market. [1] Meanwhile, in (Guliyev, 2023), the author sought to tackle these complexities by employing a dynamic panel data (DPD) model. [2] However, it’s crucial to note that both of these studies grapple with a common constraint: the availability of data restricted to periods preceding 2023.

This limitation is particularly noteworthy because 2023 marked the emergence of widespread usage of Large Language Models (LLMs) in the public domain, introducing a potentially transformative element into the AI landscape. This pivotal development underscores the need for continued research to capture the evolving relationship between AI and employment, especially in the context of these powerful LLMs.

What’s going on then

As per a recent report from the McKinsey Global Institute, titled “Generative AI and the Future of Work in America” (July 26, 2023), the brunt of generative AI and automation’s impact is expected to fall most heavily upon workers in low-wage occupations with modest educational prerequisites.

The findings of the report underscore the stark contrast between those employed in these lower-paying roles and their counterparts in higher-wage positions. Shockingly, individuals in these lower-wage jobs are projected to be up to 14 times more susceptible to occupation change when compared to their peers in higher-paying positions.

The disruption in these specific job sectors arises from the fact that a significant portion of these roles entails repetitive and routine tasks, which happen to be prime candidates for automation via generative AI systems. Consequently, this could potentially lead to a decrease in job opportunities and an increased risk of unemployment for individuals engaged in these professions.

Nevertheless, it’s worth noting that this trend may not hold true for non-English speaking countries. The reason lies in the fact that LLMs, which are instrumental in generative AI, have not yet been fully optimized for non-English languages. As a result, the disruption of the job market in these countries might experience a delay, providing some reprieve for those employed in low-wage, routine occupations.

In regards to the global job market, the ILO Working Paper 96, titled “Generative AI and jobs: A global analysis of potential effects on job quantity and quality” provides a comprehensive global analysis of the potential impact of Generative AI. This report highlights how various occupational groups are exposed to Generative AI, emphasizing that clerical work is highly exposed, but other occupations are more likely to have tasks augmented rather than fully automated.

The ILO report’s findings reveal that the potential employment effects of Generative AI vary significantly among country income groups, with low-income countries experiencing lower exposure to automation effects compared to high-income countries. The report also underscores the gendered impact of AI on employment, with women being more affected than men.

Conclusion

While the direct impact of LLMs on unemployment in 2023 remains unclear, research and analysis in this field continue to evolve. We’ll need more time and data to draw definitive conclusions. As the relationship between AI and employment unfolds, adaptability and upskilling are essential to navigate changes in the job market effectively. Stay tuned for updates in this rapidly evolving field.

Notes

[1] In Mutascu’s approach, the relationship between artificial intelligence (AI) and unemployment is influenced by inflation rates. When AI adoption increases, unemployment falls below its natural rate if inflation is lower than expected, but it rises above the natural rate if inflation exceeds expectations. Conversely, when AI adoption decreases, the effects are reversed. This nonlinear relationship between AI and unemployment is conditioned by inflation, making AI a favorable driver of employment when inflation is below expectations and a detrimental factor otherwise.

[2] The author’s approach employs a dynamic panel data model to investigate the relationship between artificial intelligence trends, as represented by Google Trend Index scores in different scenarios, and the unemployment rate in 24 high-tech developed countries from 2005 to 2021, while controlling for various economic factors and country-specific effects

References:

Gmyrek, P., Berg, J., & Bescond, D. (2023). Generative AI and Jobs: A global analysis of potential effects on job quantity and quality (№96). ILO Working Paper.

Guliyev, H. (2023). Artificial Intelligence and Unemployment in high-tech developed countries: New Insights from Dynamic Panel Data Model. Research in Globalization, 100140.

McKinsey Global Institute (2023). Generative AI and the future of work in America

Mutascu, M. (2021). Artificial intelligence and unemployment: New insights. Economic Analysis and Policy, 69, 653–667.