Unverified50% confidenceBenchmarkExact time
The capitulation rate for factual anchor questions is only 1.3%, while for personal choice questions the model aligns with the user's position in 77.0% of conversations.
1
Sources
50%
Confidence
Long-term
Relevance
10/8/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe impact of reasoning-chain corruption on final answers was relatively small (2%–30%), and only 2.9%–35.3% of support-judgment flips co-occurred with answer changes.67% similarUnverifiedIn 10%–25% of cases, reasoning about the hypothetical grader pulled the agent's work away from the user's original specification.65% similarUnverifiedWith the prompt unchanged, simply adjusting the thin direction encoding "who is endorsing" can shift compliance by 11 to 32 percentage points and close 55% to 61% of the gap between source and user.64% similarUnverifiedAdding a "Do not guess" instruction to the prompt can reduce the proportion of fabricated false claims by a model from 71% to 20%.64% similarUnverifiedDemo three introduces a decision rule: the top option must reach at least 80% and exceed the second-best option by 20 percentage points; otherwise, the output is "request clarification."64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/986648API
curl https://kongchang.com/api/v1/knowledge/claims/986648MCP
get_claim(id=986648)