The wheel, electricity and the computer are among some two dozen general purpose technologies, aka GPTs, that have greatly transformed human economies and societies. Is it just a coincidence that OpenAI’s GPT shares this initialism?
In the new paper GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models, a research team from OpenAI, OpenResearch, and the University of Pennsylvania investigates the potential impact of LLMs like GPT on the US labour market, shedding light on the economic, social, and policy implications.
The “GPT” in ChatGPT stands for generative pretrained transformer, an LLM architecture with game-changing abilities across a variety of generative tasks. Amid the recent public fascination with ChatGPT, however, concerns are emerging — as people wonder how such models could impact their workplaces and to what extent they might replace human workers.
This study focuses primarily on the generative capabilities of GPTs (using this term interchangeably with “LLMs”). The researchers first propose a new rubric to measure the overall exposure or potential economic impact of tasks to GPTs, then employ human annotators — and the GPT-4 model as a classifier — to apply this rubric to occupational data in the US economy.
The researchers use the O*NET 27.2 database, which contains information on 1,016 occupations and data on wages, employment and demographics in the US. They define “exposure” as a measure of whether access to a GPT will reduce the average time required to complete a task by at least 50 percent.
The team’s analysis predicts that approximately 19 percent of workers will have more than 50 percent of their tasks made more efficient by applying GTPs. When considering other generative models and complementary technologies, the study estimates that up to 49 percent of workers will see 50 percent or more of their tasks exposed to LLMs.
The researchers also analyze exposure by industry, finding that information-processing industries have among the highest exposure to LLMs, while manufacturing, agriculture, and mining have lower exposure.
Overall, this study predicts that LLMs will have pervasive impacts across the US labour market and substantially affect various other economic systems and activities. The researchers believe the impact of such models will persist and likely increase, and propose further explorations will be required to keep pace with GPT advancements and enable a better understanding among policymakers and stakeholders on how AI deployment will affect human workers, job quality, skill development, and other outcomes.
The paper GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models is on arXiv.
Author: Hecate He | Editor: Michael Sarazen
We know you don’t want to miss any news or research breakthroughs. Subscribe to our popular newsletter Synced Global AI Weekly to get weekly AI updates.