Context Drift
Gradual, often unnoticed divergence between what AI context describes and what is actually true about the system, project, or workflow it represents.
Category: AI
Tags: ai, context-engineering, context-management, ai-context-quality
Explanation
Context drift is the gradual misalignment between what AI context describes and what is actually true about a system, project, or workflow. Unlike context rot, which emphasizes staleness over time, context drift highlights the **directional** nature of the problem: the context slowly moves away from reality as the underlying system evolves.
Think of it like configuration drift in infrastructure. You start with a perfectly accurate description of reality. Then small changes accumulate: a function gets renamed, a convention shifts, a tool gets swapped out. Each change is minor, but the cumulative effect is a context that confidently describes a system that no longer exists in that form.
## Why It Is Dangerous
Context drift is particularly insidious because it is invisible. The AI still produces coherent-sounding output based on drifted context. There is no error, no crash -- just a slow decline in relevance. The law of staleness applies: the older the context entry, the higher the probability it has drifted.
## Mitigation
Mitigation requires treating AI context like live documentation: version-controlled, regularly reviewed, and validated against the current state. Context hygiene practices and periodic audits help catch drift before it compounds into full context rot. Context anchoring -- externalizing decisions into living documents -- provides a structural defense against drift by ensuring key context is explicit and reviewable rather than implicit and ephemeral.
Related Concepts
← Back to all concepts