This is something that I think should be studied more. I think a large reason chatbots are so popular, is that they are just used as a sounding board for people who either have no inner monologue, or no one else to talk to, or they’ve never kept a notebook/journal of any kind. The bots help them get their thoughts out, and while stroking their ego, it helps them think through problems. The llm isn’t actually solving anything. It’s just helping those people think through it.
It’s interesting to me. I think a lot of people addicted to LLM’s probably aren’t good at critical thinking, and that’s why they enjoy using it, because it guides them along (and can reinforce their preconceived beliefs, which is the dangerous part).
It’s definitely a bad thing for human brains. It will be an interesting field of study in the future for sure.
I’ve been forced to use it for my job and, yeah, you can definitely rubber duck with an AI. I’ve lost count of the times I’ve gotten halfway through a prompt and just realized what I have to do.
I’ve been subjected to raving from people that swear it’s either going to fundamentally change coding or completely make coders obsolete. And, having read these people’s code pre-AI, I’m pretty confident in saying that if you view AI coding as anything other than a fallible tool, you’re not nearly as good at your job as you think you are.
I post about 3 out of ever 10 troubleshooting questions I start to write.
“Oh, these assholes are going to ask me if I did X. Fine! I’ll fucking do it so I can tell them that I did it!” And then it works.