One of the things I found bizarre was about Charlie the chatbot—which is a great name. You said, “We found that Charlie provided accurate answers in only 2 out of the 6 questions we asked it, while the other public web-based conversational artificial intelligence tools answered 5 out of 6 questions accurately.”
How is it that the government's specialized tool for the CRA is worse than the average tool that anyone can access anywhere?