What llm isn’t a sycophant?

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...

They're trained to be sycophants as a side effect of the same reinforcement learning process that trains them to dutifully follow all user instructions. It's hard (though not impossible) to teach one without the other, especially if other related qualities like "cheerful", "agreeable", "helpful", etc. also help the AI get positive ratings during training.

I'm using ChatGPT with "Robot" personality, and I really like the style it uses. Very short and informative, no useless chatter at all.

I guess that personality is just few words in the context prompt, so probably any LLM can be tailored to any style.

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.

Kimi K2 is notably direct and free of this nonsense.