Algorithmic Bias
Updated 2026-08-02
INTRODUCTION
English translation pending.
CORE DEFINITION
Algorithmic bias refers to systematic and unfair outcomes produced by automated systems, arising from unrepresentative training data, poorly chosen objective functions, or design decisions made during development. Because models learn patterns from historical records, they reproduce and can amplify past discrimination. The problem is not limited to explicit protected attributes, since correlated variables can carry the same information. Key qualification: bias is not purely a technical defect, since the definition of fairness and the choice of objective are value judgments that must be made explicitly.
SCAFFOLDING EFFECT
Reduce cognitive load
- Audit the inputs: check whether the training data represents every group the system will affect. - Define fairness: state explicitly which fairness criterion the system is required to satisfy. - Monitor after launch: measure outcomes across groups on an ongoing and documented basis.
Anchor fast decisions
Models minimize error against historical data, so any pattern in that data, including discriminatory ones, becomes part of what the model learns. Removing protected attributes does not help when correlated features carry the same signal. The system therefore reproduces past outcomes at scale and with an appearance of objectivity that discourages scrutiny.
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/Algorithmic_biasverified
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