GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models (Eloundou, Manning, Mishkin, and Rock) investigates the potential implications of large language models (LLMs), such as Generative Pre-trained Transformers (GPTs), on the U.S. labor market.
The study finds that around 80% of the U.S. workforce could have at least 10% of their work tasks affected by the introduction of LLMs, while approximately 19% of workers may see at least 50% of their tasks impacted.
The analysis suggests that, with access to an LLM, about 15% of all worker tasks in the US could be completed significantly faster at the same level of quality. When incorporating software and tooling built on top of LLMs, this share increases to between 47 and 56% of all tasks.
The rubric used to assess occupations in this report measures the overall exposure of tasks to LLMs. It follows the spirit of prior work on quantifying exposure to machine learning and is designed to provide a framework for understanding the evolving landscape of language models and their associated technologies. The rubric employs human annotators and GPT-4 itself as a classifier to apply it to occupational data in the U.S.
The analysis indicates that most occupations exhibit some degree of exposure to LLMs, with higher-wage occupations generally presenting more tasks with high exposure. However, the report does not identify any specific industries or job types that are more likely to be impacted by the introduction of LLMs.
The report suggests that the projected effects of LLMs span all wage levels, with higher-income jobs potentially facing greater exposure to LLM capabilities and LLM-powered software. However, finds that most occupations exhibit some degree of exposure to LLMs, regardless of wage level.
The study finds that with access to an LLM, about 15% of all worker tasks in the US could be completed significantly faster at the same level of quality. When incorporating software and tooling built on top of LLMs, this share increases to between 47 and 56% of all tasks.