Artificial intelligence · Technical definition

AI system mutability.

Also known as Algorithmic mutability · Model mutability
Definition

The ability of an AI system to change its behaviour or its relevant evidence after an assessment. It may arise from model updates, configuration adjustments, new data or drift in the user population.

Nature

Operational definition

Further detail

Mutability is not necessarily a defect. It requires recording versions, defining reassessment triggers and detecting when the system in operation no longer corresponds to the system that was examined.

Cite or embed this term

Embed the definition on your site (it stays linked to the source and updates automatically), or use the glossary's open data.

Open data (JSON) →
Related signed pieces
More Artificial intelligence terms

AI Management System · AIMS

Documented system of processes, resources and responsibilities for an organisation to manage …

Algorithmic governance

Set of practices, controls, responsibilities and verification mechanisms through which an organisation…

EU AI Act

European regulation establishing prohibitions, obligations for certain high-risk systems…

NIST AI RMF

Voluntary framework from the United States National Institute of Standards and Technology for managing…

Model risk

The possibility that a model will produce unsuitable results and lead to significant consequences. Its…
Next

Apply this definition.

Define the term, see how I use it in the archive and, if you still have a question, write to me.

Define Full glossary Return to the index of technical concepts. Explore further Technical reports See the concept applied with sources. Discuss Write to me Send me a question or a specific case.
WhatsApp G-CERTI