Gradual degradation in AI model performance over time as real-world data patterns change from the original training data.
Model drift is when an AI system’s performance changes as inputs, user behaviour, or underlying assumptions shift. Even if the model is unchanged, your data and expectations are not.
Browse related definitions in the same glossary category.
AI Grounding
Techniques to anchor AI outputs in verified sources and facts, reducing hallucinations and improving reliability.
AI Hallucination
A phenomenon where a large language model generates false or illogical information but presents it as fact.
Citation Confidence
A measure of how accurately an AI model attributes information to its original source.
Content Provenance (C2PA)
Technical standards for verifying the origin and authenticity of digital content, increasingly important for AI-generated media.
Embeddings
Numerical representations of text, images, or other data that capture semantic meaning, enabling similarity search and machine learning.
Fine-Tuning
Adapting a pre-trained AI model to a specific task or domain by training it on additional specialised data.
Related GEO services, audits, and frameworks that operationalise Model Drift in commercial execution.
Understanding "Model Drift" is just the first step. Our team at TwoSquares specializes in technical SEO and digital strategy, helping brands turn complex concepts into measurable growth.