AI and Coder Employment: Compiling the Evidence [pdf] [Federal Reserve] (www.federalreserve.gov)

A recent study by Federal Reserve researchers Leland D. Crane and Paul E. Soto evaluates the impact of large language models (LLMs), particularly ChatGPT, on coder employment since its introduction in November 2022. By analyzing data from the Occupational Information Network (O*NET) linked to Current Population Survey (CPS) data, the researchers conclude that employment growth for coders has decelerated significantly, experiencing an approximate 3% annual decline compared to pre-ChatGPT levels. This slowdown is not solely attributable to broader industry trends, indicating an occupation-specific shock directly associated with the advent of advanced AI coding assistants. The findings are crucial for the AI/ML community as they highlight the nuanced effects of generative AI on labor market dynamics, particularly in highly exposed occupations like programming. The study suggests that while AI can enhance coder productivity, it may also lead to decreased demand for coders if businesses require fewer personnel to meet coding service demands. This duality underscores the need for ongoing research into how AI deployment might evolve employment structures, task distributions, and the introduction of new coding-intensive roles, as industries adapt to these transformative technologies. This research sets a foundation for better understanding the long-term implications of AI on labor markets and prompts further exploration into the complexities of AI integration in professional environments.