Chapter 09 · Sustainability & AISustainability & AI
Algorithmic accountability
Definition
Holding organisations answerable for the behaviour and impacts of the algorithmic systems they design, deploy or use — regardless of how automated those systems are.
References
This source provides part of the technical or institutional basis for the definition.
This source supports the explanation of how the term is applied, measured or governed in practice.
This source supports the wider sustainability significance and context described in the entry.
Overview
What it means
The concept responds to the 'accountability gap' created when decisions are delegated to opaque systems. ACM's 2017 statement on algorithmic transparency and accountability set early principles: awareness, explanation, auditability and responsibility sit with the deploying institution, not the algorithm.
How it is used
Mechanisms include algorithmic impact assessments, documentation duties, audit rights and liability rules. The EU AI Act, Canada's automated-decision directive and sectoral regulation each instantiate the principle differently.
Why it matters
'The algorithm decided' is not an answer. Algorithmic accountability insists that a named institution owns each automated decision — the foundation on which all other AI governance mechanisms rest.