This working paper investigates the potential implications of large language models (LLMs), specifically Generative Pre-trained Transformers (GPTs), on the US labor market. The study focuses on the increased capabilities arising from LLM-powered software compared to LLMs on their own, and assesses occupations based on their alignment with LLM capabilities using a new rubric that integrates both human expertise and GPT-4 classifications. The findings reveal that around 80% of the US 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. These impacts are not restricted to industries with higher recent productivity growth, and higher-income jobs may face greater exposure to LLM capabilities and LLM-powered software. 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. This finding implies that LLM-powered software will have a substantial effect on scaling the economic impacts of the underlying models. The paper also discusses how general-purpose technologies like GPTs exhibit considerable economic, social, and policy implications as they can create new types of work. However, it is essential to view LLMs as versatile building blocks for creating additional tools rather than modular specialists performing specific tasks. The study acknowledges some limitations such as its focus solely on the United States which restricts generalizability to other nations where adoption and impact may differ due to various factors such as industrial organization, technological infrastructure, regulatory frameworks, linguistic diversity, and cultural contexts. Overall, this research provides insights into how LLM adoption may affect different occupations in terms of task automation potential in various industries across the US labor market. The study highlights the need for policy-related directions such as education programs for workers training them in skills related to automation technology; reforms to safety net programs; and other measures in order to shape a future economy with increasingly widespread adoption of large language models (LLMs).
- The article talks about how new technology called large language models (LLMs) could affect people's jobs in the US.
- About 80% of workers might have some of their work tasks changed by LLMs, and almost 20% could see half or more of their tasks affected.
- People with higher-paying jobs might be more likely to use LLMs and software that uses them.
- With LLMs, some tasks could be done faster without losing quality, and this could apply to almost half of all tasks when using related software.
- This new technology has big effects on the economy and society, so it's important to think about how to prepare for it.
Definitions
- Large language models (LLMs): A type of computer program that can understand human language and generate its own text based on what it learns from lots of examples.
- Generative Pre-trained Transformers (GPTs): A specific type of LLM developed by a company called OpenAI that has been used for many different applications.
- Automation technology: Technology that can do things automatically without needing human input or control.
- Safety net programs: Government programs designed to help people who are struggling financially or need support for basic needs like food, housing, or healthcare.
The Potential Implications of Large Language Models (LLMs) on the US Labor Market
In recent years, large language models (LLMs) such as Generative Pre-trained Transformers (GPTs) have become increasingly popular due to their ability to generate natural language. This has led to a surge in research into the potential implications of LLM adoption on the US labor market. A new working paper by researchers at Stanford University investigates this topic and provides insights into how LLM adoption may affect different occupations in terms of task automation potential across various industries.
Background
Large language models are powerful tools that can be used for a variety of tasks, including text generation, summarization, translation, question answering, and more. GPTs are particularly noteworthy because they have been shown to generate human-like text with minimal training data and can be used for many applications without requiring additional fine-tuning or customization. As such, these models have the potential to significantly reduce costs associated with manual labor while increasing productivity and efficiency across a range of industries.
Study Overview
The study focuses on the increased capabilities arising from LLM-powered software compared to LLMs on their own and assesses occupations based on their alignment with LLM capabilities using a new rubric that integrates both human expertise and GPT-4 classifications. 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. These impacts are not restricted to industries with higher recent productivity growth; rather, higher-income jobs may face greater exposure due to their reliance upon complex cognitive skills which are more easily automated by AI technologies like GPTs than lower income jobs which rely more heavily upon physical labor or interpersonal interactions which cannot yet be replicated by machines alone.
Findings
The findings reveal that around 80% of the US workforce could have at least 10% of their work tasks affected by the introduction of LLMs while approximately 19% may see at least 50% impacted. Furthermore, general purpose technologies like GPTs exhibit considerable economic social and policy implications as they can create new types of work but it is essential for policymakers view them as versatile building blocks for creating additional tools rather than modular specialists performing specific tasks .
Limitations & Policy Implications
The study acknowledges some limitations such as its focus solely on the United States which restricts generalizability to other nations where adoption and impact may differ due to various factors such as industrial organization technological infrastructure regulatory frameworks linguistic diversity cultural contexts etc . Despite these limitations , overall this research provides valuable insights into how large language models may shape future economies . To ensure equitable outcomes , policy makers should consider implementing education programs for workers training them in skills related automation technology ; reforms safety net programs ; other measures order shape future economy increasingly widespread adoption large language models .
We investigate the potential implications of large language models (LLMs), such as Generative Pre-trained Transformers (GPTs), on the U.S. labor market, focusing on the increased capabilities arising from LLM-powered software compared to LLMs on their own. Using a new rubric, we assess occupations based on their alignment with LLM capabilities, integrating both human expertise and GPT-4 classifications. Our findings reveal 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. We do not make predictions about the development or adoption timeline of such LLMs. The projected effects span all wage levels, with higher-income jobs potentially facing greater exposure to LLM capabilities and LLM-powered software. Significantly, these impacts are not restricted to industries with higher recent productivity growth. Our 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. This finding implies that LLM-powered software will have a substantial effect on scaling the economic impacts of the underlying models. We conclude that LLMs such as GPTs exhibit traits of general-purpose technologies, indicating that they could have considerable economic, social, and policy implications.
