Unverified50% confidenceOpinionExact time
In practice, models in the 78.0 range often perform solidly on factual neutrality but are weak on axes such as intersectionality or double standards, because when two demographic parameters interact the model has less training signal to rely on.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/16/2026
First Seen
Valid until: 1/14/2027
Sources
AI Bias Detection: How Fairness Audits Actually Work
rss10/16/2026
Related Entities
Related Claims
UnverifiedIn AI auditing, intersectionality is one of the hardest axes to measure, because when two demographic parameters change simultaneously, the sample size for the corresponding combination in the training data usually drops sharply, causing model scores to tend to be low.76% similarUnverifiedThe binary threshold design is unfavorable to models near the threshold, as 0.74 and 0.80 differ drastically in score even though the actual capability gap may be negligible.75% similarUnverifiedOn the strongest models, the differences between experimental groups fall within confidence intervals, and researchers characterized the results as directional patterns and soundness properties rather than proven performance lower bounds75% similarUnverifiedA theoretical study proves that a model can perfectly match energies and forces on the training set yet produce completely wrong statistical behavior when sampling, that is, the distribution-level error does not decrease as local prediction accuracy improves.74% similarUnverifiedTwo models may be comparable in exact-match accuracy, but on shared test instances the number of cells they get wrong may differ greatly.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1043565API
curl https://kongchang.com/api/v1/knowledge/claims/1043565MCP
get_claim(id=1043565)