Audit Culture

In the AI Story

Strathern's 1997 analysis of British universities identified audit as a distinctive cultural formation — not merely the application of quantitative tools to academic work but a transformation in what academic work was understood to be. Once the measurements were in place, the measurements became the reality: scholars optimized for the metrics, the metrics optimized for legibility, and the activities the metrics could not capture progressively lost institutional support.

Berg and Seeber applied this framework to the specific dynamics of contemporary universities: how impact factors distort research agendas toward publishable results, how student evaluations pressure faculty toward entertainment over challenge, how grant funding shapes research questions toward fundable forms. The audit does not prohibit other activities — it simply makes them economically unviable.

The AI transition extends audit culture in two directions. First, AI tools generate new forms of telemetry: keystroke counts, lines-of-code metrics, tokens-generated benchmarks. The quantitative apparatus expands to cover every aspect of knowledge work that was previously invisible to measurement. Second, AI tools shape the work they measure: the worker whose performance is assessed by AI-augmented output quantity produces AI-augmented output quantity. The measurement becomes the practice.

The result is what might be called recursive audit: a regime in which the tools, the metrics, and the practices co-evolve toward configurations that maximize measurable output and make the unmeasurable progressively invisible. Berg and Seeber's warning acquires new force: in audit culture, the work that matters most becomes the work that cannot be seen.

Origin

Strathern's concept emerged from the 1990s wave of New Public Management reforms in the UK, but its theoretical scaffolding drew on Michel Foucault's work on governmentality and the specific French sociological tradition of examining how measurement constitutes rather than merely describes its objects.

Key Ideas

Measurement Constitutes. Audit culture does not merely observe existing activity — it transforms the activity into something that can be measured, and in doing so transforms its character.

Displaced Judgment. Professional judgment, which had been the governance mechanism of craft activity, is replaced by metrics designed to be legible to non-practitioners.

The Optimization Trap. Once metrics are in place, rational actors optimize for the metrics — a rationality that can produce collective outcomes nobody intended and nobody wants.

Recursive Audit in AI. AI tools simultaneously enable new forms of measurement and shape the practices they measure — closing a feedback loop in which the metric becomes the activity.

The Unmeasurable. Audit culture's deepest cost is the systematic invisibility of activities that resist measurement — depth, judgment, care, the conditions under which meaningful work becomes possible.