My $0.02 is that this a very dangerous idea with the current state of GPT. My experience with ChatGPT is that it has a tendency to make up bullshit when it doe

My $0.02 is that this a very dangerous idea with the current state of GPT. My experience with ChatGPT is that it has a tendency to make up bullshit when it doesn't have the answer to a question instead of admitting that it doesn't know, and that's an extremely bad example to set for children. Maybe some day this could happen; it probably is possible to make a language model that isn't an arrogant jerk, but it needs to be thoroughly tested first.