The AI-enabled language model technology ChatGPT has shown its abilities to do some things as well or better than people can do them. But many things might be better left to human intelligence — like accounting. That’s because some experts are finding ChatCGT is not always good at math. Paulo Shakarian, an associate professor in the School of Computing and Augmented Intelligence, part of the Fulton Schools, says accounting requires capabilities in logic that the technology does not possess. Shakarian’s tests have shown ChatGPT to be less than acceptably accurate for things like accounting, which requires translating words into mathematical equations.