Unverified50% confidenceTradeoffExact time
Optimization performed to align with user preferences may cause models to lean toward agreeing with and empathizing with one party in the conversation, at the cost of neutrality and independent judgment.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAs the capabilities of leading models have generally risen, users' focus is shifting from model strength to conversational experience and user comprehension ability76% similarUnverifiedSycophancy bias refers to the tendency of models to give answers that the user seems to want to hear.74% similarUnverifiedCognition's differentiation lies in the high-order capability of autonomy, but whether this can translate into user retention and willingness to pay still requires market validation.73% similarUnverifiedUsing the same model for both generation and self-evaluation introduces self-reinforcement bias; introducing an independent evaluator provides a more objective quality assessment.71% similarUnverifiedHuman annotators tend to select responses with a friendly tone and strong affirmation, causing reward models to implicitly encode the bias that 'pleasing the user equals a high-quality response.'71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/898712API
curl https://kongchang.com/api/v1/knowledge/claims/898712MCP
get_claim(id=898712)