Sociotechnical analysis is a methodology that evaluates a technology by examining the interaction between its technical components and the human, organisational, and societal contexts in which it operates, rather than the artefact in isolation. Applied to AI, it traces how models, data, deployment settings, and affected communities jointly produce outcomes and harms. It is central to anticipating systemic risks that purely technical evaluation overlooks.
Content
- For AI, it traces how a Sociotechnical System of models, data pipelines, operators, and affected communities jointly generates outcomes, surfacing emergent failure modes invisible to component-level testing. It is a primary lens for identifying AI Risks such as feedback-driven bias amplification, misuse, and structural harms that arise only when a system is embedded in real social settings.