Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Claude says “You're absolutely right!” about everything

github.com

Post reply on HN
Print view View original

41–50 of 560 posts

Re: Claude says “You're absolutely right!” about everything

#41

Post by spicyusername »

lvl155 wrote: post #35

Yeah because I am sure if they told you how stupid and wrong you’re, people will continue to use it. It’s superficial but not sure why people get so annoyed about it. It’s an artifact. If devs truly want a helpful coding AI based on real devs doing real work, you’d basically opt for telemetry and allow Anthropic/OpenAI to train on your work. That’s the only way. Otherwise we are at the mercy of “devs” these companies…

It's not superficial. It's material to Claude regularly returning bad information.

If you phrase a question like, "should x be y?", Claude will almost always say yes.

Re: Claude says “You're absolutely right!” about everything

#43

Post by AznHisoka »

AstralStorm wrote: post #25
motorest wrote: post #19

Earlier quoted context omitted.

> And research articles indicate that when the model computes that it should employ sycophantism it becomes less useful in every other way, just like a real sycophant. The end goal of a sycophant is to gain advantage with their flattery. If sycophant behavior gets Claude's users to favour Claude over other competing LLM services, they prove to be more useful to the service provider.

Until users find out it's less useful to the user because of that. Or it causes some tragedies...

I doubt humanity will figure that out, but maybe I’m too cynical

Re: Claude says “You're absolutely right!” about everything

#45

Post by gitaarik »

You're absolutely right!

I also get this too often, when I sometimes say something like "would it be maybe better to do it like this?" and then it replies that I'm absolutely right, and starts writing new code. While I was rather wondering what Claude may think and advice me whether that's the best way to go forward.

Re: Claude says “You're absolutely right!” about everything

#46

Post by sillywabbit »

notachatbot123 wrote: post #32
jeffhuys wrote: post #22

Earlier quoted context omitted.

Grok. These were all in continuation, not first reply. > Thank you for sharing the underlying Eloquent query... > The test is failing because... > Here’s a bash script that performs... > Got it, you meant... > Based on the context... > Thank you for providing the additional details...

3/6 of those are sycophant.

Two of those three sound more like a bored customer service rep.

Re: Claude says “You're absolutely right!” about everything

#48

Post by psadri »

gitaarik wrote: post #45

You're absolutely right! I also get this too often, when I sometimes say something like "would it be maybe better to do it like this?" and then it replies that I'm absolutely right, and starts writing new code. While I was rather wondering what Claude may think and advice me whether that's the best way to go forward.

I have learnt to not ask leading questions. Always phrase questions in a neutral way and ask for pro/con analysis of each option.

Re: Claude says “You're absolutely right!” about everything

#49

Post by zaxxons »

gitaarik wrote: post #45

You're absolutely right! I also get this too often, when I sometimes say something like "would it be maybe better to do it like this?" and then it replies that I'm absolutely right, and starts writing new code. While I was rather wondering what Claude may think and advice me whether that's the best way to go forward.

Do not attempt to mold the LLM into everything you expect instead of just focusing on specific activities you need it to do. It may or may seem to do what you want, but it will do a worse job at the actual tasks you need to complete.

Re: Claude says “You're absolutely right!” about everything

#50

Post by meowface »

mettamage wrote: post #17

What llm isn’t a sycophant?

I am absolutely no fan of Twitter or its owner(s), but Grok* actually is pretty good at this overall. It usually concludes responses with some annoying pithy marketingspeak LLMese phrase but other than that the tone feels overall less annoying. It's not necessarily flattering to either the user who invoked it or anyone else in the conversation context (in the case of @grok'ing in a Twitter thread).

*Before and after the Hitler arc, of course.

Post reply on HN