A new research paper, ChatGPT as Economics Tutor: Capabilities and Limitations, from the University of Bristol has used CORE’s The Economy 1.0 as a reference text. The paper’s aim is to assess the ability of ChatGPT to explain economic concepts and answer multiple-choice questions.
The paper, authored by Natalie Bröse, Christian Spielmann, and Christian Tode, found that newer GPT models generate very accurate responses, but some inaccuracies persist. CORE’s The Economy 1.0 was used as reference material to determine the accuracy of GPT-3.5, GPT-4o, and o1preview model’s responses to the book’s definitions and solutions. The paper explains the decision to use CORE’s text on page 6 and how this choice impacts results on page 28. It is argued that ChatGPT can serve as an effective automated tutor for basic, knowledge-based questions—supporting students while posing a relatively low risk of misinformation.
Read the full paper ChatGPT as Economics Tutor: Capabilities and Limitations on the University of Bristol website.