We Have Never Been Modern
We Have Never Been Modern (1991; English translation 1993) is the book that most sharply articulated Latour's challenge to the modern constitution. Across 150 compact pages, it argues that the modernity Western societies believe themselves to inhabit has never actually existed. What exists instead is a constant proliferation of hybrids — entities simultaneously natural and social — paired with a rhetorical operation that insists hybrids do not exist, that each hybrid is really one pure thing or the other. The book reframes the crisis of postmodernism not as the collapse of modernity but as the recognition that modernity was always a self-misdescription. For AI, the analysis is foundational: if the modern constitution cannot accommodate hybrids, and AI is paradigmatically hybrid, then the governance architectures that AI demands cannot be produced by the purified categories modern institutions have inherited.
The Persistence of Institutional Power — Contrarian ^ Opus
There is a parallel reading that begins from the lived experience of those navigating AI deployment in actual organizations. While Latour's analysis elegantly diagnoses why modern institutions fail to govern hybrids, it underestimates how successfully those same institutions have captured and domesticated hybrid phenomena when their power is threatened. The modern constitution may be a fiction, but it is a fiction backed by centuries of accumulated institutional authority, legal precedent, and economic control. When AI emerges as a hybrid that threatens existing arrangements, the response is not institutional paralysis but aggressive colonization—tech companies become utilities, algorithms become trade secrets, and machine learning becomes another domain of regulatory compliance.
The proliferation of hybrids that Latour identifies has not, in practice, undermined the modern settlement between Nature and Society. Instead, it has produced a vast apparatus of translation that converts every hybrid threat into terms the existing institutions can digest. Climate change becomes carbon markets; pandemics become pharmaceutical patents; AI becomes either technical standards or ethical guidelines, never both simultaneously. The practitioners who work with AI daily—engineers writing code, lawyers drafting terms of service, workers whose jobs are being automated—experience not the collapse of modern categories but their violent reimposition. The hybrid nature of AI is acknowledged precisely to the extent that it can be parsed into manageable regulatory buckets. What remains unaddressed is not the hybrid character of the technology but the political economy of who controls it. The modern constitution persists not because we believe in it but because those who benefit from the division between technical and political authority have every incentive to maintain it, even as they privately acknowledge its fiction.
In the AI Story
The book emerged from Latour's reading of the 1660 Boyle-Hobbes debate, which Steven Shapin and Simon Schaffer had reconstructed in Leviathan and the Air-Pump (1985). Boyle won the right to speak for Nature through experimental demonstration; Hobbes won the right to speak for Society through sovereign authority. The settlement, Latour argued, constituted the modern division of knowledge and power. Scientists got Nature; politicians got Society; and the proliferation of hybrids — entities that refused the division — was rendered invisible by the rhetorical work of purification.
The book's central claim is that modernity is defined by a constitutional contradiction. Its official self-description insists on strict separation between nature and society, fact and value, non-human and human. Its actual practice proliferates hybrids constantly — greenhouse gases, GMOs, AIDS, nuclear reactors — that cut across every division. The contradiction is stable because the two operations (proliferation and purification) are performed by different institutions that do not talk to each other. Scientists proliferate hybrids in laboratories while denying that they are doing so. Politicians regulate hybrids in assemblies while denying that the technical details matter to political deliberation. The system works by mutual misrecognition.
The consequence is that the problems most characteristic of the contemporary moment — climate change, pandemics, ecological collapse, AI — are precisely the problems modernity cannot address. They are hybrids. Addressing them requires governance architectures that take hybridity seriously — that recognize the simultaneous natural and social character of the phenomena in question and deliberate accordingly. The existing institutions, organized around the purified categories, keep failing not because they are badly designed but because they are designed for a world that does not exist.
For the AI moment, the book's diagnosis is precise. AI is simultaneously technical artifact and social institution, mathematical optimization and political instrument, individual tool and civilizational transformation. The existing governance architectures — technical regulation on one side, civic advocacy on the other — reproduce the purified division and thereby fail to address the hybrid. The book does not solve the problem of AI governance. It identifies why the problem is hard: the existing categories cannot describe what needs to be governed, and building new categories is the prerequisite for any institutional response adequate to what AI actually is.
Origin
The book was published in French in 1991 and translated into English by Catherine Porter in 1993. It emerged from Latour's engagement with ecological politics in the late 1980s and his participation in debates about what French intellectuals called 'postmodernity.' His position was idiosyncratic: rather than celebrating or lamenting the end of modernity, he argued that modernity had never been what it claimed to be, and that the 'postmodern' condition was merely the moment at which the self-misdescription became harder to sustain.
The book was received with controversy in both anglophone philosophy (which found its style unfamiliar and its arguments against realism alarming) and French sociology (which found its rejection of Bourdieu's critical framework provocative). Over the following decades, as climate change and biotechnology made hybrid phenomena unavoidable, the book's influence spread across disciplines that had initially resisted it.
Key Ideas
Modernity as self-misdescription. The modern constitution never accurately described Western practice. It was always a rhetorical operation that proliferated hybrids while denying that hybrids existed.
The Boyle-Hobbes settlement. The division of labor between experimental science and sovereign politics was a contingent historical settlement, not a natural or necessary arrangement.
Proliferation and purification. Modernity's two simultaneous operations: producing hybrids and rhetorically denying them. The stability of the constitution depends on keeping the operations separate.
Postmodernity as recognition, not rupture. The postmodern condition is not the collapse of modernity but the moment at which the constitutional contradiction becomes visible.
A nonmodern future. The path forward is neither to preserve modernity nor to abandon it, but to acknowledge the hybrids and build governance structures adequate to what the world has always actually contained.
Appears in the Orange Pill Cycle
The Uneven Dissolution of Categories — Arbitrator ^ Opus
The tension between Latour's diagnostic clarity and the contrarian's institutional realism resolves differently depending on which layer of the AI phenomenon we examine. At the conceptual level, Latour is entirely correct (100%): AI is irreducibly hybrid, and attempts to govern it through purified categories of 'technical' or 'social' consistently fail to capture what makes it transformative. The modern constitution's categories simply cannot describe what AI is doing to work, meaning, creativity, and knowledge. Every attempt to reduce AI to either mathematical optimization or social impact misses the phenomenon itself.
At the level of institutional response, however, the contrarian view dominates (80%): existing power structures have proven remarkably adept at metabolizing hybrid threats without fundamentally restructuring. The EU's AI Act, OpenAI's safety theater, and corporate AI ethics boards all demonstrate how institutions can acknowledge hybridity rhetorically while maintaining operational separation between technical and political governance. The modern constitution bends but does not break. Where both views converge is in recognizing that this containment is temporary—the contrarian admits that violent reimposition is required precisely because the categories are failing.
The synthetic frame the topic requires is temporal: we are living through the extended moment of modern category failure that Latour diagnosed, but institutional inertia means this failure plays out over decades, not years. AI governance is simultaneously impossible (Latour is right about the conceptual inadequacy) and inevitable (institutions will impose inadequate frameworks anyway). The real work happens in the gap between these truths—in the specific practices, workarounds, and hybrid forums that emerge when practitioners must govern something their categories cannot describe. The dissolution of the modern constitution is uneven, contested, and generational, not revolutionary.
Further reading
- Bruno Latour, We Have Never Been Modern, trans. Catherine Porter (Harvard University Press, 1993)
- Steven Shapin and Simon Schaffer, Leviathan and the Air-Pump (Princeton University Press, 1985)
- Bruno Latour, Politics of Nature (Harvard University Press, 2004)
- Isabelle Stengers, Cosmopolitics I (University of Minnesota Press, 2010)
- Philippe Descola, Beyond Nature and Culture (University of Chicago Press, 2013)