Recently, I had two wildly different encounters with ChatGPT—both within a single day. Right before lunch, I asked it to write some code in Python—the ubiquitous computer programming language—to carry out a complicated data analysis. It took just a few seconds to produce an elegant and (perhaps more important) correct solution. Before ChatGPT, I would have needed at least an hour to code it up on my own. A drastic improvement in productivity, to put it lightly.
That evening, I checked on my daughter, who was feverishly studying for an Advanced Placement European history exam the next day. That exam, she explained, would require her to draw a map of Europe, complete with every country’s boundary and capital city, from memory.
Out of curiosity, I put that same task to ChatGPT. In a few seconds it drew a map that correctly outlined the borders of France and identified Paris as its capital. Same for Hungary and Budapest, Athens and Greece, and so forth. But it also identified “Beme” as the capital of “Germanyn,” and “Limerace” as the capital of “Bublin.” And according to ChatGPT, there’s a country called “Vienna” with “Benin” as its capital. Who knew? Here I’d give ChatGPT an A for productivity, but a C+ for reliability.
What do the fictional cities of Limerace, Bublin, and Beme, Germanyn have to do with government finance? Quite a lot, it turns out. In fact, here’s a bold prediction: in three years, most local government chief finance officer (CFO) job descriptions will include “data integrity assurance for artificial intelligence applications.”
- Publication date: June 2025
- Author: Justin Marlowe