Have you noticed how AI feels amazing and super smart in areas which you don't know much about. It sounds so confident, knowledgeable, and incredibly convincing. It's impossible to not trust it.
But once you start asking questions about the topics where you yourself are an expert -- you immediately see where it falls short and how the generated response is mediocre at best. Sometimes even border-edge harmful.
But the LLM itself never doubts in its responses. It's not designed to sound like a person who's aware of their own limitations. It never expresses that in the output. The confidence level is always dialed to 11.
Keep that in mind.
Especially when you use AI as your tax advisor 😉
