ai::semantic_drift
Semantic Drift
When the same term acquires different meanings across systems, teams or AI agents.
Canonical Definition
The silent divergence of meaning that occurs when the same term acquires different definitions across teams, systems, models or AI agents, producing inconsistent decisions without raising any system error.
Alternative Interpretations
Why the same term causes confusion across teams.
Machine Learning
Statistical drift in input distribution over time (data drift).
Linguistics
Historical change in word meaning across decades or centuries.
Enterprise AI
Same term carrying different operational meanings across systems and agents today.
Business Impact
What happens when different meanings of Semantic Drift are used in parallel.
Operational Risk
Two teams reach opposite decisions from 'the same' input field.
AI Risk
Multi-agent workflows produce contradictory outputs from identical prompts.
Compliance Risk
Decisions are not reproducible against a single authoritative meaning.
Integration Risk
APIs labelled identically return semantically incompatible payloads.
Related Entities
- ai::semantic_drift → opposed by → ai::alignment
- ai::semantic_drift → solved by → governance::definition_control
- ai::semantic_drift → resolved through → enterprise::system_of_record
Governance Metadata
This definition is governed. Not merely documented.
- Owner
- Head of AI Governance
- Status
- aligned
- Version
- 1.0
- Effective Date
- 2025-02-15
- Source
- WikiSure AI Working Group
- Domain
- AI