Like the Romans used to drink from lead cups. Something like that.
(it appears that the moderator is removing any replies with which he disagrees. It might be a piefed issue. But all the removed replies are top level and upopular. Stranger coincidences have happened I suppose.)
In that case, you are asking it a question you don’t know the answer to, and the answer could be wrong. You might have the knowledge to know when the output isn’t factual and reprompt it, or you may not always know when the output isn’t factual; the potential detriment to you probably isn’t so much cognitive, it’s potentially wasting your time, money, supplies and labor on an incorrect answer, or worse, leaving the customer to deal with bad service or worksmanship from an incorrect answer.
Sure, but the starting point is that I already don’t know something and need to look it up one way or another. Yes, the AI answer might be wrong but the guy on the YouTube video or online forum could equally be wrong. That is not something unique to AI.
I’m also obviously not just blindly trusting what it says. When it recommends me a product for example I don’t just go buy it without a question. The AI is just pointing me to the hopefully right direction for me to figure it out but the ultimate decision still remains mine.
A good recent example:
“I need to paint plywood that gets wet regularly, should I use paint A or paint B?”
“Neither, you should use paint C”
*Looks up paint C*
“Yes, I definitely should use paint C”