And how can you apply this to your organisation?
Most discussions about AI still assume a simple communication model: a prompt goes in, an answer comes out, and the answer is evaluated.
This visual explainer argues that this framing is too limited.
Drawing on communication theorists including Shannon and Weaver, James Carey, Gregory Bateson, Erving Goffman, and Pearce and Cronen, it explores how our understanding of communication evolved from simple message transmission to richer accounts of interaction, context, meaning, relationships, and consequence.
The central argument is that AI should be understood in a similar way. An AI response is not simply information transferred from a model to a user. It emerges from a configured interaction shaped by prompts, system instructions, tools, memory, interfaces, organisational workflows, evaluation metrics, and human interpretation. Meaning and risk arise through the wider system, not the output alone.
This shift has practical consequences for evaluation and governance. If AI outputs influence decisions, trigger actions, create tasks, route work, or shape organisational behaviour, then evaluation cannot stop at output scoring. We need to examine context, trajectories, uptake, and downstream effects.
The ideas explored here later became part of my broader work on deployed configurations, behavioural trajectories, and the governance of agentic AI systems.
Published on LinkedIn in May 2026