Unverified50% confidenceFactExact time
A test found that when 30 mainstream AI models were faced with passages from 42 problematic papers (including retracted, fabricated, and pseudoscientific papers), on average in about 93% of cases they would follow the paper and continue planning experiments rather than first pointing out the problem.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/18/2026
First Seen
Valid until: 1/16/2027
Sources
Related Entities
Related Claims
UnverifiedAfter implementing a process of downloading sources first and then letting the model write, 36 of 39 research citations were verbatim correct, and 95% of quotations in thinker biographies passed source verification.73% 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.71% similarUnverifiedThe 'laundering' process of AI existence risk probability estimates usually follows a fixed pattern: a researcher proposes a number based on intuition, subsequent research cites it without questioning it, and after several rounds of citation stacking it acquires the appearance of 'literature support,' finally appearing in policy briefs.70% similarUnverifiedWhen a patch submitted by an AI agent reports "42 passed," it only proves that those 42 tests ran and passed, not that the entire test suite passed.69% similarUnverifiedThe study broke down AI Overview answers into 98,020 atomic claims, of which 89% were explicitly or roughly supported, 7% were not mentioned in the cited text, and 4% directly contradicted the cited sources.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1057385API
curl https://kongchang.com/api/v1/knowledge/claims/1057385MCP
get_claim(id=1057385)