Black Box Problem
Updated 2026-08-10
INTRODUCTION
English translation pending.
CORE DEFINITION
The black box problem refers to systems, particularly deep neural networks, whose internal decision process cannot be reconstructed or explained even though their inputs and outputs are observable. This creates difficulties for debugging, accountability, and regulatory compliance, especially in high-stakes domains such as lending, hiring, and medicine. The field of explainable artificial intelligence has developed methods such as feature attribution, surrogate models, and probing to recover partial explanations. Key qualification: opacity is a matter of degree, since many black boxes can be explained locally.
SCAFFOLDING EFFECT
Reduce cognitive load
- Demand an account: ask who can explain a decision that affects someone materially. - Choose the tool: match the explanation method to the stakes of the decision being made. - Keep a fallback: retain a human review path where a full explanation is required.
Anchor fast decisions
Deep models distribute a decision across millions of interacting parameters, so no single weight corresponds to a human-readable reason. Because the mapping is nonlinear and highly distributed, the same input pattern can be processed through many different internal routes. Explanations therefore approximate behavior locally rather than describing the actual computation.
MINIMUM ACTION
In progress 0/1Practice this model in one real situation:
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more
Source support: Explicit
- en.wikipedia.orghttps://en.wikipedia.org/wiki/Explainable_artificial_intelligenceverified
PRIVATE NOTES · Only visible to you
SAVED Q&A
ENTRY Q&A · Private saving available
Ask with a clear boundary
thinkingmodels answers from published entry context only.
Your question is sent to thinkingmodels. The answer uses public entry context only.
RELATED MODELS