Like many other people, I have enjoyed giving prompts to ChatGPT, an AI chatbot created by OpenAI. ChatGPT will respond to a wide range of prompts, often in amusing ways, yet often in ways that are surprisingly impressive. 

I have seen many of my fellow academics express concern that students will use this type of AI to cheat. I can understand their concern. The AI wrote some material that I could imagine grading favorably. For example, at one point I asked the AI to write a Christopher J. Coyne article. It did not write an article Chris would write. However, it did write a fairly good summary of his excellent book After War: The Political Economy of Exporting Democracy

However, as OpenAI directly warns users, ChatGPT “may occasionally generate incorrect information.” This is certainly true on economic topics. Every time I have asked ChatGPT about the Alchian-Allen effect, the AI has talked about different economic phenomena, never about the Alchian-Allen effect itself. I imagine the same is true for at least some other economic concepts.

Students should be careful about using ChatGPT as a substitute for real research and studying. They certainly should not treat it as though it’s “better than Google” for finding factual information. Instructors concerned about students using ChatGPT to cheat should try directly asking ChatGPT to answer the same questions they ask students. The instructors might be pleasantly surprised by how ineffective cheating using ChatGPT would be. Or they may choose to drop questions that the AI answers too reliably.

Does ChatGPT have implications for economic education beyond the academic integrity concerns it raises? Perhaps! I have been having fun asking ChatGPT to write song lyrics. So far I’ve had it write songs about a variety of economic topics, including microeconomics, public choice, and Elinor Ostrom. These types of whimsical prompts could offer educators new and memorable ways to present material, though students might find the results more cringeworthy than appealing. 

That said, faculty should also be careful to fact check any educational materials they try to create using ChatGPT. Just as the AI’s errors could cause trouble for unscrupulous students, they might embarrass us too if we’re not careful!

 


Nathan P. Goodman is a Postdoctoral Fellow in the Department of Economics at New York University. His research interests include defense and peace economics, self-governance, public choice, institutional analysis, and Austrian economics.