Scaling Laws for Economic Productivity: Experimental Evidence in LLM-Assisted Translation

TL;DR

This study establishes empirical scaling laws linking Large Language Model compute to economic productivity improvements, demonstrating significant gains in translation performance and earnings, especially for lower-skilled workers, with implications for U.S. economic growth.

Contribution

It provides the first experimental evidence of scaling laws for economic outcomes in LLM-assisted translation, quantifying productivity gains from increased model compute.

Findings

A tenfold increase in compute improves translation speed by 12.3%.

Model scaling boosts earnings per minute by 16.1%.

Lower-skilled workers benefit four times more from scaling.

Abstract

This paper derives "scaling laws"--empirical relationships between the training compute of Large Language Models (LLMs) and their performance--for economic outcomes. In a preregistered online experiment, 300 professional translators completed 1,800 tasks using one of 13 LLMs (or a control). A tenfold increase in model compute improved task completion speed by 12.3%, grades by 0.18 standard deviations, and earnings per minute by 16.1%. Gains were four times larger for lower-skilled workers. These findings suggest continued model scaling could boost U.S. productivity by at least 6.9% over the next decade.