GPTs are GPTs: an early look at the potential impact of large language models on the labor market
The study “GPTs are GPTs: An Early Look at the Potential Impact of Large Language Models on the Labor Market” by Tyna Eloundou, Sam Manning, Pamela Mishkin, and Daniel Rock explores the potential implications of large-scale language models (LLMs), such as pre-trained generative transformers (GPTs), on the US labor market. The study focuses on the enhanced capabilities resulting from LLM-driven software compared to LLMs alone.
The research team developed a new classification system to evaluate occupations based on their alignment with LLM capabilities. This system integrated both human experience and GPT-4 ratings. The findings revealed that around 80% of the American workforce could see at least 10% of their work tasks affected by the introduction of LLM. Likewise, around 19% of workers could see at least 50% of their work tasks impacted.
The authors make no predictions about the development or timing of LLM adoption. The projected effects span all wage levels, with higher-income jobs potentially facing greater exposure to the capabilities of LLMs and LLM-based software. Significantly, these impacts are not limited to industries with higher recent productivity growth.
The analysis suggests that with access to an LLM, about 15% of all worker tasks in the US could be completed significantly faster while maintaining the same level of quality. When LLM-based software and tools are incorporated, this proportion increases to between 47% and 56% of all tasks. This finding implies that LLM-driven software will have a substantial effect on the extent of the economic impacts of the underlying models.
The study concludes that LLMs such as GPTs exhibit characteristics of general-purpose technologies, indicating that they could have considerable economic, social and political implications. The findings underscore the need for a deeper understanding and careful regulation of these emerging technologies to maximize their benefits and minimize potential harm to the labor market.
I have written this article simply using the content of the document: “GPTs are GPTS”.