Default to surrender
AI chatbots highlight a challenge that is worth understanding. It applies to customer service, bureaucracies and teachers as well…
If you ask an AI a question and it’s not confident in the answer, it should say, “I’m not sure.”
That could be followed up with, “do you want me to guess” or “if you could give me more context…”
But proudly and confidently bluffing isn’t helpful.
It’s really unusual to find this sort of humanity (even with humans), but it’s a great way to build trust.