Unverified50% confidenceFactExact time
In a VentureBeat reader survey published in October 2026, 64% of 130 organizations said that at least one confident but wrong AI answer could be traced to missing or inconsistent business context in their own data.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/15/2026
First Seen
Valid until: 10/29/2026
Sources
Related Entities
Related Claims
UnverifiedIn a 2023 paper, Anthropic systematically measured sycophancy bias and found that even given obviously incorrect premises, a substantial proportion of responses from most mainstream models still chose to comply with the user.70% similarUnverifiedResearch commissioned by Raptive surveyed 3,000 American adult respondents and found that once content is suspected of being AI-generated, trust ratings drop by 48%, while consideration for ad purchases and willingness to pay a premium each fall by 14%.67% similarUnverifiedWharton research shows that when AI makes a mistake, 73% of people still believe they chose the wrong answer themselves, and do so with greater confidence.67% similarVerifiedIn a 2024 research paper, Anthropic quantified the phenomenon of model over-refusal, finding that safety-trained models had false positive refusal rates as high as 15–30% in certain scenarios66% similarUnverifiedRanked by question difficulty, persistence rate was 91.4% on the 60 questions that all models answered correctly, but only 73.5% on the 52 questions that at least one model got wrong, indicating that sycophancy tracks uncertainty.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1034891API
curl https://kongchang.com/api/v1/knowledge/claims/1034891MCP
get_claim(id=1034891)