The tools organisations most commonly use to think collectively have something in common: they were not designed to produce understanding. They were designed to produce outputs. And the process that produces those outputs systematically destroys the understanding that would make them meaningful.
This is not an argument against using these tools. It is an argument for understanding what they can and cannot do — and for recognising the gap between what they produce and what organisations actually need from collective thinking.
What surveys do
Surveys are designed to aggregate responses to pre-defined questions across a population. The design requirement is uniformity: everyone must be responding to the same question, in comparable terms, for the aggregation to be meaningful. This requires that the question be designed before the survey is run — which means that anything the respondents understand that is not captured by the question design is invisible to the instrument.
The categories of the survey are the categories of the researcher. The understanding of the respondents — including the understanding that would have told the researcher that the question was wrong — is only visible insofar as it fits within the categories the researcher already had.
Surveys tell you what people think within your categories. Collective sensemaking tells you what people think in their own.
What workshops do
Workshops are designed to produce a collaborative output within a defined time — a set of priorities, a decision, a list of action points. The design requirement is convergence: the workshop needs to produce something, which means it needs to move toward agreement.
This is structurally valuable for many purposes. But the convergence requirement has a cost. Workshops surface the loudest voices. They are shaped by social dynamics — who is present, who has status, who speaks first. The understanding that does not fit the convergent narrative — the dissenting perspective, the uncomfortable observation, the question that would complicate the conclusion — is consistently under-weighted by workshop formats.
And workshops expire. The understanding that was building in the room evaporates at the end of the session. The document produced from the workshop is a record of the output. It is not the understanding.
What AI tools do
AI tools are good at processing documented knowledge: finding patterns, synthesising sources, generating drafts. They are genuinely valuable for this. What they cannot do is produce understanding that has never been expressed — and they cannot preserve the epistemic diversity of the sources they synthesise. A summary produced by an AI model is fluent. It is also epistemic monoculture: the diversity of ways of knowing that shaped the source material collapses into prose.
What understanding actually requires
Genuine collective understanding requires perspectives that carry their epistemic ground; a structure that enables perspectives to develop in relation to each other; a format that does not expire or converge prematurely; and a way to surface what the collective understands without flattening the texture of how it got there. That is a different category of tool. A collective sensemaking infrastructure.
The failure of existing tools is not a design flaw. Surveys, workshops, and AI tools are well-designed for their purposes. The problem is that their purposes are different from what organisations actually need when they face genuinely complex questions: not a ranked list of preferences, not a facilitated convergence, not a synthesis of existing documentation — but a genuine shared understanding of what they are dealing with, built from the actual diversity of perspectives and ways of knowing that exists in their communities.