Now a days people refer to AI instantly for an answer. Maybe it’s a problem they could have solved within 10min in their head or a question they could have done an internet search on and read a few forums to figure out. However, now people go straight to AI which is known to give many many wrong answers.
I know a couple people that fit this bill and now I almost completely disregard what they say. We’ll even be talking face to face and they’ll ask AI something from our conversation in real time.
It depends what they ask.
A lot of questions are fairly accurate to ask AI.
Also, some people are smart enough to habe configured their pre-prompt to reduce common issues with AI platforms.
There’s a big difference between:
“Hey whats the answer to (esoteric question)”
Vs
“Hey can you find me a paper on (well documented topic)” (and then opening the paper up and reading the actual source)
Using it as a powerful fuzzy search tool with semantic intent is fine.
Its the difference between asking a librarian how to cure cancer, vs asking the librarian to help you quickly find a book on cancer treatments.
But isn’t that just a good search engine we should have separate to a LLM at that point? Semantic search doesn’t need a full blown chat not in between.