As I said, absolutely unethical. Even if the error percentage was the same or less than that of a human, which I really don’t believe given the human inputs were training data so even at 100% the LLM would be as flawed as a human, then still in any industry where decisions are being made which impact people like accounting, forensics, logistics etc then having a machine making all of those decisions without contextual awareness is a disaster waiting to happen.

But yeah you don’t have to convince me that fortune 500 CEOs are cheap assholes cutting costs, I believe you there.