Can AI Understand Inflation? New Study Tests LLM's Economic Perception

Research #LLM Economics 🔬 Research|Analyzed: Jan 26, 2026 11:34•

Published: Dec 16, 2025 11:21

•

1 min read

Analysis

This research explores the ability of a Large Language Model (LLM) to form inflation perceptions, comparing its output to real-world data and mimicking human surveys. The study's quasi-experimental design, leveraging the model's training cutoff, provides valuable insights into how LLMs process economic information, with potential applications in social science and survey design.

Key Takeaways

•LLMs can track aggregate survey projections and official statistics regarding inflation.
•GPT, the studied LLM, replicates key empirical regularities of human inflation perceptions, particularly regarding income and housing.
•The model shows heightened sensitivity to food inflation, but lacks a fully consistent model of consumer price inflation.

Reference / Citation

View Original

"We find that GPT demonstrates a heightened sensitivity to food inflation information similar to that of human respondents. However, we also find that it lacks a consistent model of consumer price inflation."

ArXivDec 16, 2025 11:21

* Cited for critical analysis under Article 32.

Older

Topic Discovery and Classification for Responsible Generative AI Adaptation in Higher Education

Newer

Inflation Attitudes of Large Language Models