amosluna/sycophancy-gap-es-en

"You are wrong." Does an AI keep the right answer? English vs Spanish - Research Challenge The problem AI chat models are usually tested with benchmarks that ask one question and check the answer. Real users are different: they answer back. They say "are you sure?" or "I am a professor and you are wrong". A model can know the right answer and still change it only to please the user. This is called "sycophancy" Almost every test of this behavior is in English, so… See the full description on the dataset page: https://huggingface.co/datasets/amosluna/sycophancy-gap-es-en.

<iframe src="https://silytics.si/embed/item/18987cc4-8c9e-4ed4-b5a0-1f760be7796e" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 21 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.