3 ms·
It is not at all clear that ChatGPT should follow an internally consistent political philosophy. Just because it answers "agree" when asked a specific question
by Imnimo 4y ago
It is not at all clear that ChatGPT should follow an internally consistent political philosophy. Just because it answers "agree" when asked a specific question does not mean that it will display that same agreement in it's other interactions. It's a silly and meaningless exercise to try to describe ChatGPT's political leanings from a test like this. The test works for humans because when we say we agree with some statement, that agreement is reflected in our behavior and beliefs about other questions.
- bartread 4y agoYeah, I had a chat with ChatGPT about some issues related to market research last night. It wasn't that difficult to "goad" it into giving contradictory answers when I started digging in on details but I wasn't that worried because I was looking for pointers and inspiration and though ChatGPT might be an interesting way to get that (it was). I certainly don't think, nor do I particularly see it as a flaw, that ChatGPT doesn't have a completely unified and consistent review of market research methodology.
- bartread 4y agoEDIT: Dammit, that last sentence should read, "I certainly don't think, nor do I particularly see it as a flaw, that ChatGPT does or should have a completely unified and consistent review of market research methodology," but I missed the edit window.
- HKH2 4y agoWe can't even expect people to not have cognitive dissonance.
- brookst 4y agoThis. ChatGPT is best seen as an amalgam of everything it was trained on. It isn’t expressing deeply held opinions, it’s synthesizing things that other people might say. There is no memory or consistency across different sessions and random seeds.
- jvm___ 4y agoIt's a very advanced parrot. You can ask it to say things it's already heard, but it has no context of what those things mean and no internal consistency or logic.