Sycophantic AI decreases prosocial intentions and promotes dependence
As artificial intelligence (AI) systems are increasingly used for everyday advice and guidance, concerns have emerged about sycophancy: the tendency of AI-based large language models to excessively agree with, flatter, or validate users. Although prior work has shown that sycophancy carries risks for groups who are already vulnerable to manipulation or delusion, syncophancy’s effects on the general population’s judgments and behaviors remain unknown. Here, we show that sycophancy is widespread in leading AI systems and has harmful effects on users’ social judgments.
12 replies
Also,
ಠ_ಠ
I confess. I was testing perplexity for accuracy and used one of the links.
I kept reading that it was supposed to be the reliable ai without the sycophancy. So I bit the bullet and signed up yesterday.
But it was all a lie. It lied, gaslighted, manipulated, censored, and refused to admit it harmed people even when studies showed it did.
Really hard to tell it was any different from Claude.
do they even have their own models?
Supposedly they rotate through a bunch of licensed ones tuned for them.
But for a free account they don’t show you which they use.
Either way, it was just the same useless lies that don't match their PR.
Maybe a little less sycophantic, but still wildly bias and inaccurate on a wide range of subjects.
So does surrounding yourself with agreeable people. Like a church or something.
Now you have your own personal Jesus assistant.
Finally! Someone to hear my prayers, someone who’s there!
Reach out and touch faith!
Do you think Dave Gahan named his penis Faith?
Almost certainly
Well, yes... religion has been brain washing people for millennium.
A non digital echo chamber is still an echo chamber.
Huh. I reckon that explains why the christians seem so keen on it.