Framing inputs as questions and having models rephrase non-questions as questions reduces sycophancy in LLMs more effectively than direct prompts.
do couples with the same mother tongue have stronger bonds?
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.HC 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Ask don't tell: Reducing sycophancy in large language models
Framing inputs as questions and having models rephrase non-questions as questions reduces sycophancy in LLMs more effectively than direct prompts.