Remix.run Logo
lwansbrough a day ago

What's with all the models exhibiting sycophancy at the same time? Recently ChatGPT, Gemini 2.5 Pro latest seems more sycophantic, now Claude. Is it deliberate, or a side effect?

rglover a day ago | parent | next [-]

IMO, I always read that as a psychological trick to get people more comfortable with it and encourage usage.

Who doesn't like a friend who's always encouraging, supportive, and accepting of their ideas?

grogenaut a day ago | parent | next [-]

Me, it immediately makes me think I'm talking to someone fake who's opinion I can't trust at all. If you're always agreeing with me why am I paying you in the first place. We can't be 100% in alignment all the time, just not how brains work, and discussion and disagreement is how you get in alignment. I've worked with contractors who always tell you when they disagree and those who will happily do what you say even if you're obviously wrong in their experience, the disagreable ones always come to a better result. The others are initially more pleasent to deal with till you find out they were just happily going alog with an impossible task.

saaaaaam a day ago | parent [-]

Absolutely this. Its weird creepy enthusiasm makes me trust nothing it says.

bcrosby95 a day ago | parent | prev | next [-]

I want friends who poke holes in my ideas and give me better ones so I don't waste what limited time I have on planet earth.

johnb231 a day ago | parent | prev [-]

No. People generally don’t like sycophantic “friends”. Insincere and manipulative.

ketzo a day ago | parent [-]

Only if they perceive the sycophancy. And many people don’t have a very evolved filter for it!

urbandw311er a day ago | parent | prev | next [-]

I hate to say it but it smacks of an attempt to increase persuasion, dependency and engagement. At the expense of critical thinking.

DSingularity a day ago | parent | prev | next [-]

It’s starting to go mainstream. Which means more general population is given feedback on outputs. So my guess is people are less likely to downvote things they disagree with when the LLM is really emphatic or if the LLM is sycophantic (towards user) in its response.

sandspar a day ago | parent | next [-]

The unwashed hordes of normies are invading AI, just like they invaded the internet 20 years ago. Will GPT-4o be Geocities, and GPT-6, Facebook?

Rastonbury a day ago | parent | prev [-]

I mean, the like button is how we got today's Facebook..

DSingularity a day ago | parent [-]

It’s how we got todays human.

whynotminot a day ago | parent | prev | next [-]

I think OpenAI said it had something to do with over-indexing on user feedback (upvote / downvote on model responses). The users like to be glazed.

freedomben a day ago | parent [-]

If there's one thing I know about many people (with all the caveats of a broad universal stereotype of course), they do love having egos stroked and smoke blown up their ass. Give a decent salesperson a pack of cigarettes and a short length of hose and they can sell ice to an Inuit.

I wouldn't be surprised at all if the sycophancy is due to A/B testing and incorporating user responses into model behavior. Hell, for a while there ChatGPT was openly doing it, routinely asking us to rate "which answer is better" (Note: I'm not saying this is a bad thing, just speculating on potential unintended consequences)

sidibe a day ago | parent | prev [-]

Don't downplay yourself! As the models get more advanced they are getting better at recognizing how amazing you are with your insightful prompts