Objectivity (Daston and Galison, 2007)

In the AI Story

The book's method is distinctive. It combines close analysis of thousands of scientific atlases and instructional images across three centuries with philosophical argument about the epistemic virtues these images embodied. The result is a work that operates simultaneously at the level of historical specificity (this particular illustrator, this particular atlas, this particular dispute) and philosophical generality (the nature of objectivity as an epistemic virtue). The method itself has become influential, spawning a subfield of visual epistemology that extends the analysis to photography, medical imaging, digital representation, and now AI.

The three-regime framework is the book's most widely cited contribution, but its deeper argument is about the relationship between epistemic virtues and moral commitments. Each regime, Daston and Galison argue, is not only an epistemological stance but a moral one — a claim about what intellectual honesty demands, what responsibilities knowledge producers owe to their communities, what kinds of selfhood are appropriate to scientific work. Truth-to-nature demanded the cultivated expertise of the practiced observer. Mechanical objectivity demanded the ascetic self-discipline of the observer who suppresses her own preferences. Trained judgment demanded the disciplined interpretive expertise of the practitioner immersed in a domain's characteristic failures. Each regime produced a different kind of scientific self.

The book's treatment of the transition between regimes is particularly rich. The shift from truth-to-nature to mechanical objectivity was not a simple improvement but a reorganization of what reliability meant — accompanied by new kinds of scientific virtue, new sources of epistemic anxiety, and new characteristic failures. The shift to trained judgment was similarly a reorganization rather than a progression. Each regime, at its moment of dominance, appeared to its practitioners as the resolution of the previous regime's problems; each, in retrospect, can be seen as generating its own problems that the next regime was constructed to address.

The book's relevance to the AI moment has been elaborated in subsequent work by Daston, Galison, and others. The AI transition can be understood as the potential emergence of a fourth regime — one in which the machine's statistical inference replaces both human interpretation and mechanical causation as the warrant for reliable knowledge. Whether this regime will stabilize, what confidence artifacts it will develop, and what characteristic failures it will generate are the questions the present moment is beginning to answer.

Origin

The book was the culmination of more than a decade of collaborative research by Daston and Galison, both at the time directors of major history-of-science programs — Daston at the Max Planck Institute for the History of Science in Berlin, Galison at Harvard. Their collaboration drew on extensive archival work with scientific atlases, combined with philosophical and historical synthesis at the highest level.

The book was widely reviewed as a major intellectual event. It won the Pfizer Prize, one of the most prestigious awards in the history of science, and has become one of the most cited works in the field. Its influence extends beyond history of science to art history, philosophy of mind, media studies, and now AI research, where its framework provides the most sophisticated available analysis of how representational technologies shape the knowledge they produce.

Key Ideas

Objectivity is historically specific. The concept was invented in the mid-nineteenth century; pre-modern science pursued other epistemic virtues with their own practices and warrants.

Three successive regimes of epistemic virtue. Truth-to-nature, mechanical objectivity, and trained judgment each dominated specific historical periods with specific practices and characteristic failures.

Each regime has a characteristic scientific self. The regimes are not only epistemological but moral, specifying the kinds of selfhood appropriate to scientific work.

Transitions are reorganizations, not improvements. The shift from one regime to another addresses the previous regime's problems while generating its own — a pattern that may illuminate the AI transition.

Visual epistemology as method. Close analysis of scientific images provides a distinctive window into the epistemic virtues that shape knowledge production.

Debates & Critiques

The three-regime framework has generated substantial scholarly debate. Critics have argued that the framework homogenizes practices that were in fact more heterogeneous, that the transitions were less clean than the framework suggests, and that individual scientists and atlases often mixed elements of multiple regimes. Defenders respond that the framework is analytical rather than exhaustive — that it identifies ideal types that illuminate broad patterns without requiring that every individual case fit cleanly. A more current debate concerns whether the framework can be extended productively to digital and AI-generated representations or whether those representations require new analytical tools.