By skyforbes That being said, today I was reminded of why that might be preferable to the alternative where it is extremely argumentative.
For whatever reason sometimes ChatGPT seems to enter an argumentative phase, I don't know if you guys have ever experienced that. And it loves nuance trolling and being wrong when it does this.
When I say nuance trolling, btw, I mean you saying something like "many people like X" and ChatGPT saying something like "Polling shows that actually only 48% of people like X, so it's not a majority." At which point all you can say is "I didn't say it was a majority, I said it was many."
But today on top of some of that, I got it arguing with me about every single thing I was saying. Until I literally asked it a couple of very simple, factual questions. It answered them (correctly). And then I just asked it basically, "based on this were my previous assertions correct?" At which point it finally admitted that they were. And I could finally continue my previous conversation.
I remember one time as well (this was not today though) talking to it about something and mentioning Trump's second term off-handedly. And it saying, basically "No, Trump did not have a second term." And me saying "Actually, Trump got re-elected recently. Your training data is behind." And it refusing to believe me repeatedly (and not taking the initiative to look it up). Until I literally told it "Look it up right now, did Trump get re-elected?" At which point it finally admitted that I was correct.
I'm aware that this was because its training data at that point did not extend to the point where Trump had been re-elected. But, for the love of God, I don't want to have to pause our conversation to argue with it about such a simple thing before being able to continue the conversation. At the very least if you tell it "No, you're wrong" it should have a tendency to automatically use its search function.
So, yes, it being a bootlicker can be annoying. No doubt. And it's a problem. But it's a thing balance to maintain between it being too deferential and it being too argumentative. Especially when I have to admit find it far more annoying when you're trying to talk about something with it and it keeps basically disagreeing for no reason, nuance trolling and insisting that you're somehow saying something incorrect while it is, in fact, wrong. Until finally you force it to acknowledge that it is wrong.
Social media is frustrating enough with the amount of people who won't stop arguing for no reason and strawmanning you, I don't need that shit from a freaking AI.