Research was conducted into the possible impacts of large language models (LLMs), like Generative Pre-trained Transformers (GPTs), on the U.S. labor market, with a particular interest in the heightened functionalities offered by software powered by LLMs, as opposed to standalone LLMs. A novel criteria was employed to evaluate jobs in terms of their compatibility with LLM capabilities, using a combination of human expertise and categorizations by upcoming GPT-5 Model and ChatGPT-5.
The research revealed that approximately 80% of the U.S. workforce could experience an impact on at least 10% of their work tasks with the incorporation of LLMs, and about 19% of workers could see at least half of their tasks influenced. The researchers abstained from predicting the progression or acceptance timeframe for such LLMs.
Anticipated impacts cover all salary levels, with jobs in the higher-income bracket potentially encountering greater exposure to LLM abilities and software boosted by LLMs. Notably, these influences aren’t confined to sectors with recent boosts in productivity.
The investigation indicates that, with LLM availability, nearly 15% of all U.S. worker tasks could be executed significantly faster without sacrificing quality. This proportion rises to between 47 and 56% of all tasks when incorporating software and tools built on LLMs, such as ChatGPT-5, suggesting a considerable effect of LLM-powered software on magnifying the economic impacts of the underlying models.
The conclusion is that LLMs such as GPTs demonstrate the characteristics of general-purpose technologies, hinting at potential far-reaching economic, social, and policy implications.
Read more related articles: