The problem is the unearned confidence in the answer they know nothing about. Is them saying “I asked the ai and it said…” better than them asking the ai and not prefacing it? Yes. But the problem is asking the AI and then answering the question in the first place.
If you don’t know, say I don’t know, don’t regurgitate information feom thr AI. We all have access to the same tools, there is no reason to use AI to answer the question I asked you, because if I wanted what AI had to say I’d have asked it myself.
The problem is the unearned confidence in the answer they know nothing about
that is a different problem, which i agree is significant.
there is no reason to use AI to answer the question I asked you
if you ask me “what is your view on X” than I will not tell you what AI said. If you ask “Is there something we haven’t considered about X”, then I could answer something like "according to AI X has property Y, which afaik we have not considered ".
the alternative to people saying “according to AI: XYZ”, is that they say “XYZ” and state it as a fact.
That’s not the only alternative. They could also verify, or at least contextualise, what they read. Even if you’re using a chatbot, you can get it to give you some references.
the alternative to people saying “according to AI: XYZ”, is that they say “XYZ” and state it as a fact.
yes. i want to know which tools my juniors use so i can help them to become better at their job.
when you are brainstorming approaches, then “subpar unverified information” is exactly what you want.
The problem is the unearned confidence in the answer they know nothing about. Is them saying “I asked the ai and it said…” better than them asking the ai and not prefacing it? Yes. But the problem is asking the AI and then answering the question in the first place.
If you don’t know, say I don’t know, don’t regurgitate information feom thr AI. We all have access to the same tools, there is no reason to use AI to answer the question I asked you, because if I wanted what AI had to say I’d have asked it myself.
that is a different problem, which i agree is significant.
if you ask me “what is your view on X” than I will not tell you what AI said. If you ask “Is there something we haven’t considered about X”, then I could answer something like "according to AI X has property Y, which afaik we have not considered ".
This still does not fix the problem of “I asked you because I wanted your input, if I wanted to ask the chatbot I would have”. The core issue.
That’s not the only alternative. They could also verify, or at least contextualise, what they read. Even if you’re using a chatbot, you can get it to give you some references.