Annoyed ChatGPT users complain about bot’s relentlessly positive tone

Users complain of new "sycophancy" streak where ChatGPT thinks everything is brilliant.

See full article...

 

[Apple fanboy 101]

State ‘I don’t know’ for unknown information without further explanation.

This is the most important to me. An assistant that is unable to recognize its own limitations is just bad.

 

[Apple fanboy 101]

It's difficult what's profound and what is not. When you look at the average general knowledge, especially in the States, people are morons who can't even find their own country on a map. If that's how ignorant most people are, why can't we call that profound? Similarly, someone with good general knowledge probably knows stuff that would have sounded extremely profound to savants from 300 years ago. How do we measure what the average knowledge should be, when that knowledge always increases? Do we take the average population knowledge?

 

[Apple fanboy 101]

About a week ago I felt compelled to ask ChatGPT about this behavior. It gave a decent explanation, and then went right back to fluffing (emphasis mine).



I do enjoy some positive reinforcement when it's called for but I'm not asking for it here. I prefer subject-matter-expert Chat over BFF Chat all day. I'll be adding some of the custom instructions mentioned above.

At the same time ChatGPT will lecture you for trying to make politically incorrect jokes. And by politically incorrect I simply mean jokes about height for example. Ask it to make a joke about tall people and it will. About small people it used to lecture you, but I just tried it and it seems they’ve scaled down the political correctness about every single thing.

 

[Apple fanboy 101]

That speaks to the point Altman was actually making, which was a reference to Roko’s basilisk

This theory is the perfect example of how intelligent people can believe the stupidest thing. Because the theory itself doesn’t make any sense.