Unverified50% confidenceFactExact time
A new analysis finds that most numerical errors in AI-written papers are misattributions—citing values from the wrong experiment, wrong metric, or wrong run in one's own experimental records—rather than deliberate fabrication.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe evaluation results contained a total of 41 failure cases, of which 3 were genuine AI agent faults, 28 were bugs in the grader or test cases, 9 were reply checks that were too strict, and 1 was an email masking defect.75% similarUnverifiedAbout four-fifths of the unsupported claims corresponded to numbers present in the agent's own files, but from the wrong experiment, the wrong metric, or the wrong run batch.74% similarUnverifiedThe error rate of manual data entry is typically between 0.5% and 4%, and the error rate is higher for tasks involving complex judgment.73% similarUnverifiedIn manually proofread samples, the character error rate was only 0.68% when the text was clearly legible, but approximately 45% of sampled passages were completely uncorrectable due to source quality issues.70% similarUnverifiedOn a test set of 100 tickets, SupportOps-AI had an escalation error rate of 18%, a hallucination rate of 6%, an infinite-loop rate of 3%, and a complete failure/unresolved rate of 4%.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114260API
curl https://kongchang.com/api/v1/knowledge/claims/1114260MCP
get_claim(id=1114260)