I've noticed that ChatGPT (whatever model the free version uses by default) likes to phrase answers as though it's correcting me, even when my question doesn't contain any assumptions.
I've noticed that ChatGPT (whatever model the free version uses by default) likes to phrase answers as though it's correcting me, even when my question doesn't contain any assumptions.
Something I've noticed quite a bit on the paid plans as well is that it starts its answers with "I mostly agree..." or "almost correct...," then goes through the list of points I made without actually disagreeing with any of them.
I assumed this is a system prompt or RL that nudges it to be always skeptical but then it still has like all the models the urge to appease the user.
Interestingly, I've been noticing almost the opposite issue. 5.6 Sol frequently starts its responses with "Yes" even when my prompt doesn't contain a yes-or-no question.
I've been noticing the same thing since 5.4- it starts with "Yes" yet I haven't asked a question. I think it might be related to the reasoning, like it's answering its own questions?
It's been happening for years, and it does seem to be entirely related to its background reasoning leaking out of the context and into the output.
The "It's not X, it's Y"-style repetition is another example of that: It argues with itself in the background, and the argument leaks out as if in refutation of things that you (the human) never had imparted into discussion at all.
Doesn’t just mean the model has picked up on what happens when two graybeards who still wear cargo shorts meet and one utters a declarative sentence? :-)
One thing I despise about ChatGPT is how it goes on 4 paragraph tangents about how slight details are wrong, and then always ends with a paragraph of bolded fucking text rephrasing some statement in my original question, but with a huge amount of hedging to exclude minute counterexamples.