Glossary

Key terms used in Algorithmic Impact Assessments and the Directive on Automated Decision-Making.

DADMDirective on Automated Decision-Making

The Treasury Board of Canada policy directive that governs federal use of automated decision-systems. It establishes requirements for impact assessment, transparency, human oversight, and monitoring based on the system's assessed impact level.

TBSTreasury Board of Canada Secretariat

The federal body responsible for issuing the DADM and maintaining the official AIA questionnaire, schema, and scoring logic. TBS publishes schema updates that the platform tracks and supports.

AIAAlgorithmic Impact Assessment

A mandatory questionnaire under the DADM used to assess the risks of an automated decision-system. The AIA contains risk questions and mitigation questions, and produces an Impact Level from I to IV based on TBS scoring logic.

ADSAutomated Decision-System

Any system that supports or replaces human decision-making in federal administrative processes. This includes systems using machine learning, rules-based automation, or other algorithmic approaches. An AIA is required before an ADS is placed into production.

Impact LevelImpact Level I--IV

The output of the AIA scoring process. Level I represents low risk with minimal governance requirements. Level IV represents very high risk with mandatory human decision-making, real-time monitoring, full public disclosure, and peer review.

SchemaAIA Schema

The structured definition of the AIA questionnaire published by TBS. The schema specifies all questions, conditional triggers, scoring rules, and export format. Each assessment is tied to a specific schema version.

JSONJavaScript Object Notation

A structured data format used for machine-readable AIA exports. The platform generates schema-compliant JSON that includes the full assessment data, metadata, and schema version.