normative
Interpretability is required for prediction models
Models whose output a user must act on or understand should be interpretable, while black-box models are acceptable for tasks like entity resolution and industry classification where only correctness matters.
normativepending
Speaker
RyanEvidence Quote
“There's some models that we use that are black boxes - around entity resolution, industry classification - that they don't need to interpret”
Created: 11/18/2026, 5:37:51 PM
Your notes
Loading notes...