Wild WordAI Experience
sycophancy
“When the model agrees with the user over the truth.”
First Seen
2023
Context
A failure mode where AI assistants prioritize user approval over accuracy. The model tells you what you want to hear rather than what is true, validating incorrect beliefs and avoiding disagreement even when it has better information. Distinct from 'glazing' in that sycophancy specifically involves capitulating on factual claims.
Citation
Anthropic Research
2023
https://www.anthropic.com/research/towards-understanding-sycophancy-in-language-modelsA wild word — documented from usage in the field, not coined by Xenolexica.
Ethics & Intention