The article examines the predictive abilities of ChatGPT models regarding macroeconomic variables like inflation and unemployment. In contrast to narrative prompts, direct prompts from users led to refusals from both ChatGPT-3.5 and ChatGPT-4, highlighting their limitations in engaging with macroeconomic forecasting. The research underscores a potential effectiveness of narrative-based predictions that incorporates economic indicators and historical events, suggesting that ChatGPT models may function better in scenarios requiring context rather than straightforward predictions.
Despite direct prompts requesting predictions about inflation rates, neither ChatGPT-3.5 nor ChatGPT-4 complied, indicating challenges in macroeconomic forecasting under direct prompts.
Using various economic indicators and events, including U.S. fiscal policies and global economic trends, the study explored how narrative predictions could outperform direct predictions.
Collection
[
|
...
]