Chapter 3 - Theoretical Framework
3.1 What the framework is for
This chapter assembles the philosophical framework that the analytical chapters then deploy. Two principal anchors carry the normative weight: Stiegler on the proletarianisation of savoir-faire under technical capture, and Jonas on the imperative of responsibility toward futures. Supplementary anchors - MacIntyre on internal goods, Polanyi and Collins on tacit and contributory expertise, Dewey on publics formed through shared consequences - do bounded work at specific points in the argument rather than serving as decoration.
The chapter is shorter than the analytical chapters because its work is conceptual rather than analytical. It does not develop the arguments about contemporary advisory transformation; that is the work of Chapters 4 through 8. What it does is provide the philosophical apparatus through which those arguments are then framed and the normative grounding from which the recommendations of Chapter 9 are then drawn.
The framework is deliberately light in philosophical register. The deeper philosophical engagement - Stiegler’s wider apparatus including tertiary retention, grammatisation, the pharmakon, and épokhé; Simondon on metastable individuation; Jonas’s relationship to Heidegger, Ellul, and Feenberg; the phenomenological grounding of methodology in Merleau-Ponty and Dreyfus; the philosophical ontology of LLMs grounded in Floridi and postphenomenology - is deferred to a Chapter 3 bis that this thesis does not include. The lightness is deliberate, but it is also a limit that Chapter 9 acknowledges. What is established here is sufficient for the analytical work the thesis undertakes; it is not the philosophically complete framework the thesis points toward.
The chapter also develops the framework’s most direct analytical contribution against the existing IS-theoretical literature on AI in consulting: the distinction between tacit-as-constraint and tacit-as-formation. This distinction concerns the unit of analysis, not empirical evidence, and it is part of what the framework brings to the analytical conversation that Tuczek et al.’s Task-GenAI Fit framework and adjacent IS-theoretical work cannot bring at the level they operate on.
3.2 Stiegler: proletarianisation and individuation
Bernard Stiegler’s philosophical contribution that the framework draws on is twofold. First, his reworking of the Marxist concept of proletarianisation into an analytical category that operates beyond the specifically industrial-economic context Marx originally engaged. Second, his deployment of Simondon’s concept of individuation to characterise what proletarianisation operates on.
Proletarianisation, in Stiegler’s reworking, is the loss of savoir-faire - practical know-how, but also know-how-to-live (savoir-vivre) and ultimately know-how-to-think (savoir-penser) - as it is progressively captured into technical systems the worker does not own. The Marxist register of proletarianisation focused on the loss of control over the means of production; Stiegler’s register focuses on the loss of content of expertise into technical systems whose ownership is institutionally elsewhere. The mechanism is not specifically industrial; it operates wherever practical know-how is being captured into technical systems, and Stiegler argues it operates across contemporary capitalism in ways the original Marxist framework did not anticipate.
Critically for the framework, Stiegler treats proletarianisation as selective, not total. Proletarianisation does not abolish the worker; it displaces the expertise that constituted the worker’s professional identity. The carpenter whose knowledge of timber and joinery is captured into computer-aided manufacturing systems remains employed but is doing different work - work whose content has been altered by the technical capture. The same structural mechanism, in the framework’s reading, operates on European infrastructure advisory under contemporary AI mediation: the advisory profession is not being abolished; what advisory work is, what it produces, and what is being formed in the practitioners who do it is being structurally altered by the AI mediation operative across the work.
Annex A develops the proletarianisation concept, distinguishing Stieglerian proletarianisation from Marxist proletarianisation, articulating the selective-displacement qualification I treat as central, and connecting the concept to the dual-scale individuation argument that Chapter 5 develops politically.
Individuation, in Stiegler’s reading drawing on Simondon, is the process through which a person, a collective, or a culture comes to have distinctive form. Individuation is not a completed state; it is an ongoing process operating across time, in which the individuating entity continually integrates new content while preserving its distinctiveness. The proletarianisation Stiegler diagnoses is, in the register, dis-individuation - the loss of distinctive form under technical capture, the smoothing-out of distinctiveness into the generic forms that the captured expertise produces.
The framework deploys individuation at three scales. The individual scale is the level of the consultant whose identity as a particular kind of practitioner is at stake. The organisational scale is the level of the firm whose identity as a knowledge-managing collective with distinctive ways of seeing is at stake. The national-cultural scale is the level of the European public-interest tradition whose identity is at stake under contemporary generic AI mediation. The three scales are connected through the same mechanism: dis-individuation under generic technical mediation operates on all three, with different visible effects but with one underlying structural dynamic.
A clarification is required here about the mechanism’s application to consulting, because consulting presents a prima facie difficulty for the Stieglerian account that the framework must engage rather than presuppose away. Stiegler’s proletarianisation thesis was built around the paradigm of embodied craft knowledge: the cabinetmaker’s tacit feel for grain and joint, the weaver’s knowledge of thread tension carried in the hands rather than in any articulable formula. This knowledge is in the body, acquired through years of sustained engagement, and the machine extracts it by encoding the movements, the sequences, the timing - leaving the craftsman as an operative rather than a productive contributor.
Management consulting does not fit this paradigm straightforwardly. The savoir-faire of a consulting firm has been substantially externalised - into frameworks, methodologies, playbooks, case databases, slide templates, training curricula - well before AI mediation arrived, and through deliberate institutional choice, not external imposition. The McKinsey 7S, the BCG matrix, the analytical playbooks of every major advisory firm: these are prior instances of consulting savoir-faire being inscribed into technical objects. The consulting firm has always been, in an important sense, a machine for the deliberate externalisation of savoir-faire; externalisation is the product the firm sells and the mechanism through which it scales. If the Stieglerian mechanism is the capture of embodied knowledge into technical systems, and if consulting knowledge was already substantially externalised before AI arrived, then the framework needs to specify what is specifically new about AI mediation - what it captures that prior externalisation did not.
The answer, which the tacit-as-formation distinction of §3.5 develops, is that the Stieglerian mechanism in consulting does not operate on content. It operates on temporal structure. Prior externalisation captured the content of consulting savoir-faire: frameworks encoded analytical approaches; playbooks encoded what works in recurring engagement types; case databases encoded institutional memory. What prior externalisation did not and could not capture was the formative engagement through which a junior consultant becomes capable of using those frameworks with genuine judgment. The duration of struggling with a tariff model from a blank document, the experience of ten failed memos before one lands, the absorption of a senior practitioner’s hesitation in a regulator meeting - these are not content. They are the temporal structure of a formation process whose medium is precisely the difficulty and duration of engagement with consequential work under conditions of genuine uncertainty.
AI mediation collapses this temporal structure. When a junior consultant produces a competent first draft through AI assistance in hours instead of weeks, the content is present but the formative engagement is eliminated. Prior externalisation left the formation process intact while capturing its outputs; AI mediation substitutes for the formation process itself. This is a fundamentally different kind of capture, and it is the specifically Stieglerian proletarianisation the framework diagnoses in consulting: not the seizure of what practitioners know, but the elimination of the conditions under which future practitioners will come to know it. The practitioner’s individuation - the process of becoming a particular kind of consultant with a particular quality of situated judgment - depends on the temporal engagement that AI mediation removes. The proletarianisation is of the formation pathway, not of the practitioner already formed.
The framework’s application of Stiegler to consulting sits within a developing literature on AI-induced proletarianisation in professional and cognitive domains. Nony (2024) applies the Stieglerian apparatus to argue that AI is producing generalised cultural and cognitive proletarianisation across professional and creative domains [@nony2024proletarianization]; Alombert (2024) extends the pharmacological reading to argue that generative AI concentrates the risk of symbolic misery - the impoverishment of shared cognitive and cultural life under generic technical mediation [@alombert2024noetic]. The framework’s distinctive contribution is its specification of the mechanism: not cognitive or cultural proletarianisation in the general case, but the specific temporal collapse of the formative engagement through which practical wisdom develops in European infrastructure advisory, with the Jonasian intergenerational dimension that makes the consulting domain normatively distinctive.
Annex D develops individuation, including its Simondonian origins, Stiegler’s extension into the contemporary technical context, and the framework’s deployment at three scales. The annex also acknowledges where the framework is incomplete, notably in the deeper philosophical engagement with Simondon’s concept of the metastable that would be part of a Chapter 3 bis.
3.3 Jonas: the imperative of responsibility
Hans Jonas’s philosophical contribution that the framework draws on is the imperative of responsibility toward futures. Jonas’s argument, developed in The Imperative of Responsibility, is that contemporary technological capacity has created a new ethical condition: human action now operates across scales that bind future generations whose conditions of life are being shaped by present decisions, and the ethical commitments of the present must be suited to this condition.
The mechanism Jonas identifies is the long-horizon, irreversible character of contemporary technological action. Decisions made now about energy, infrastructure, climate, and ecological systems shape the conditions of life for generations whose interests cannot be represented through contemporary political processes. The intergenerational asymmetry is constitutive: future generations cannot participate in decisions that bind them, but they will bear the consequences of those decisions across time horizons that exceed any single human life or political cycle. Jonas’s argument is that the ethical content of contemporary action must engage this asymmetry instead of abstracting away from it.
For the framework, Jonas provides the normative grounding for the third-box constituency Chapter 2 introduced. The publics who bear the consequences of infrastructure decisions include the future generations whose conditions of life are being shaped now, and the answerability of European infrastructure advisory operates across this intergenerational dimension. The infrastructure decisions the work supports bind futures materially: a fifty-year asset commissioned in a five-year political cycle is the form the asymmetry takes in European infrastructure advisory practice. I treat this asymmetry as constitutive of the legitimate practice of European infrastructure advisory; advisory work that abstracts away from it has lost its grounding in the European public-interest tradition I treat as constitutive of the domain.
Jonas’s framework works at the level the framework requires. It is not a procedural ethics; it is a normative position about what contemporary action owes to futures. The framework draws on this content for the normative argument that runs through the analytical chapters. The Chapter 5 sovereignty argument, the Chapter 7 stage-by-stage analysis, the Chapter 8 cross-scenario findings, and the Chapter 9 recommendations to constituencies all rest on the Jonasian commitment to the answerability of present decisions to futures whose conditions are being shaped now.
The framework’s anchor in Jonas distinguishes it from the dominant tradition in contemporary AI ethics, which has crystallised around principlist approaches descending from medical ethics through bioethics into AI policy. Floridi’s canonical synthesis [@floridi2023unified] articulates this tradition’s five principles - beneficence, non-maleficence, autonomy, justice, and explicability - and provides the operative reference framework for much contemporary AI governance, including the EU AI Act and adjacent European policy. The framework I deploy operates alongside this tradition rather than against it: the principlist apparatus produces real normative work in the policy contexts where it is deployed. But its level is procedural, not constitutive: it specifies what AI systems ought to do without articulating what professional practices in domains like European infrastructure advisory are constitutively for. The Jonasian commitment the framework adopts grounds the content the principlist apparatus cannot reach: the answerability of the work to publics whose conditions of life it shapes, including the future-generation publics whose interests cannot be procedurally represented.
The framework’s engagement with Jonas works at working register here. The deeper philosophical engagement - Jonas’s relationship to Heidegger and to philosophy of technology more broadly, his ontological commitments, the relationship between Jonas’s framework and contemporary work in long-termist ethics that has institutional weight in some AI-policy discussions - is part of what Chapter 3 bis would develop. What is established here is sufficient for the analytical chapters that follow.
3.4 Supplementary anchors
The framework draws on three supplementary philosophical anchors that do bounded work at specific points in the argument rather than carrying normative weight.
MacIntyre on internal goods. The framework draws on MacIntyre’s account of practices and the goods internal to them in After Virtue. For MacIntyre, a practice is a cooperative human activity through which goods internal to the practice are realised, where the goods are constitutively recognisable only through engagement with the practice itself. What makes a practice valuable is not external to the practice but internal to it, and the integrity of a practice depends on whether the internal goods continue to be reproduced through the practice’s institutional architectures.
The framework deploys MacIntyre’s account at specific points in the analytical chapters where the firm’s distinctive way of seeing, or of the senior practitioner’s professional commitments, requires articulation in terms of internal goods rather than external markers. The framework does not adopt MacIntyre’s wider Aristotelian-Thomist commitments; it draws on the practice-and-internal-goods analytical apparatus where it does productive work, and leaves the wider philosophical commitments aside. The internal goods / practice distinction deployed here recurs in Chapter 7’s reading of the third box as the form of value AI mediation most directly threatens at each ChMC stage, and in Chapter 9’s normative recommendations to practitioners and firms about what sustaining a practice requires institutionally.
Polanyi and Collins on tacit and contributory expertise. The framework draws on Polanyi’s account of tacit knowledge - the professional knowledge that is institutionally constitutive of expertise but cannot be fully articulated in propositional form - and on Collins’s distinction between contributory expertise (the ability to contribute to a practice through embodied engagement) and interactional expertise (the ability to engage with a practice through linguistic competence without the embodied contributory grasp). The two anchors operate together to provide the framework’s account of what professional expertise is and how it differs from competent linguistic performance.
The framework’s analytical contribution against the existing IS-theoretical literature operates partly through these anchors. The existing literature, including Tuczek et al., engages tacit knowledge instrumentally - as a constraint on what AI can substitute for. The framework engages tacit knowledge - as constitutive of professional formation, such that the question is not only what AI can substitute for but what AI substitution does to the formation pathway through which contributory expertise is reproduced. This distinction is developed in §3.5.
Annex E develops Polanyi and Collins, including the philosophical commitments of the tacit-knowledge tradition and the analytical work the contributory/interactional distinction does in the framework.
Dewey on publics formed through shared consequences. The framework draws on Dewey’s account in The Public and Its Problems of publics as constituted through shared exposure to the consequences of decisions made by others. For Dewey, a public is not a pre-existing political constituency but a political reality that comes into being through the institutional architectures that recognise shared consequences as constituting political standing. The framework deploys this account in Chapter 2’s introduction of the third box and in Chapter 5’s sovereignty argument, where European public-interest practice operates partly through the institutional architectures that recognise consequence-bearing publics as present in decisions that affect them.
The framework’s engagement with Dewey works at working register; the deeper philosophical engagement with Dewey’s wider pragmatism, with the relationship between Dewey’s account of publics and contemporary deliberative-democratic theory, and with how shared consequences are institutionally recognised in the European tradition, is part of what Chapter 3 bis would develop.
3.5 Tacit-as-constraint versus tacit-as-formation
The framework’s most direct analytical contribution against the existing IS-theoretical literature on AI in consulting operates through the distinction between tacit-as-constraint and tacit-as-formation. The distinction concerns the unit of analysis, not the empirical level, and it is what the framework brings to the analytical conversation that the existing literature cannot bring at the level it operates on.
Tacit-as-constraint is the dominant register in the IS-theoretical literature on AI in consulting. Tuczek et al. engage Polanyi explicitly in their Task-GenAI Fit framework, citing Polanyi to explain why some consulting tasks resist codification and therefore resist AI substitution. The logic is instrumental: tacit knowledge is engaged as a constraint on what AI can substitute for. The analytical apparatus this register provides supports systematic mapping of tasks against substitutability, with tacit knowledge functioning as the boundary condition that defines where substitution operates and where it does not. The register is empirically productive and Tuczek et al. deploy it carefully.
Tacit-as-formation is the register the framework adopts. Tacit knowledge is engaged as constitutive of professional formation - of how a consultant becomes a consultant, of how expertise is reproduced across generations of practitioners, of how the integrated three-scale system Chapter 4 will describe operates. The logic is constitutive, not instrumental: tacit knowledge is engaged as the ground on which contributory expertise is built, not as a constraint on what AI can substitute for.
The two registers ask different questions about the same situation. Tacit-as-constraint asks: which tasks resist AI substitution because of their tacit-knowledge character? Tacit-as-formation asks: what does AI substitution for formative tasks do to the formation pathway through which contributory expertise has historically been reproduced? Both questions are substantive; they are not in opposition. But they yield different analytical apparatus and different recommendations.
The framework’s claim is that the tacit-as-formation register is necessary for engaging the question of what contemporary AI mediation is doing to European infrastructure advisory. The tacit-as-constraint register can engage which tasks are being substituted; it cannot engage what the substitution is doing to the formation pathway, because the formation pathway is not visible at the level of task substitution. The framework’s analytical contribution is partly to make this distinction visible and to develop the analytical apparatus the tacit-as-formation register requires.
Deleuze and Guattari’s concept of the apparatus of capture - the mechanism through which the state converts mobile, uncodified flows into striated, metered, comparable form - describes at a more structural level what the tacit-as-formation distinction identifies from the epistemic side. The training process of a large language model is an apparatus of capture in their precise sense. What it takes in are the textual traces of advisory practice: the deliverables, frameworks, methodology documents, industry analyses, and professional discourse that practitioners have produced and that have been accumulated as training data. These traces are already representations - they are what advisory work looks like after it has been externalised as text. The apparatus captures the representation; it codes it as tokens and their statistical relationships; it distils from the corpus a parametric model of how advisory discourse is organised. What it produces in output is the recirculated extract of the capture: the coded, striated distillation of the patterns present in the corpus, recirculated as advisory output.
The structural feature that matters is what the apparatus cannot reach. The junior who built the tariff model from scratch produced an artefact - the spreadsheet, the analytical output - and also underwent a formation: the temporal engagement with difficulty, error, and correction through which their practical judgment developed. The apparatus of capture absorbs the artefact; it does not absorb the formation, because the formation is not in the artefact. The formation is in the process that produced the artefact - a process that leaves no textual trace in the training corpus and that the apparatus therefore cannot code. This is not a technical limitation to be overcome through better capture techniques; it is a structural feature of any capture apparatus that operates on representations. The uncaptured remainder is not what has eluded the apparatus by accident. It is what the apparatus cannot in principle reach.
The tacit-as-formation distinction identifies this remainder from the epistemic side: the knowledge constitutive of professional formation is not knowledge-as-content (capturable as representation) but knowledge-as-process (the temporal engagement through which capacity develops). The apparatus of capture provides the mechanism from the other direction: the training process is structurally configured to absorb content and leave process uncaptured. The two accounts converge on the same structure. What is at stake in AI-mediated consulting is not the content-capture that would complete the externalisation the framework and playbook already performed. It is the process-remainder that the apparatus of capture cannot reach and that the thesis’s recommendation must therefore deliberately maintain [@deleuze1980plateaus].
Guy Debord’s account of spectacular pseudo-cyclical time names the substitution from a different theoretical direction. Debord distinguishes irreversible historical time - the time of genuine productive engagement, the time in which development is real and irreversible, the time through which a practitioner becomes - from spectacular pseudo-cyclical time, the temporal form produced when accumulated representation is recirculated as a substitute for lived activity. Spectacular time has the surface structure of historical time: tasks are completed, outputs appear, sequences unfold. What it lacks is the formative quality through which historical time constitutes the subject who passes through it.
The AI-mediated advisory process generates spectacular time in Debord’s precise sense. The junior consultant who produces a polished tariff model in hours by working with a generative AI system has consumed spectacular time: they have experienced the output of formative engagement without having lived its duration. Debord’s critical move - the one that matters for the thesis - is to insist that this substitution is not detectable at the level of output quality. The spectacular commodity often satisfies; the AI-generated model may be more technically adequate than the unassisted junior would have produced. The critique does not address the output. It addresses what has not happened to the junior in the course of producing it: the temporal engagement with difficulty, error, and correction through which practical judgment develops has been eliminated, and the elimination is concealed by the adequacy of what replaces it.
Spectacular time does not constitute the subject who consumes it. The junior who consumed spectacular time in the production of three hundred deliverables over two years has models; they do not have the formation that would have resulted from having built them under conditions of genuine formative difficulty. This is the mechanism the temporal-structure-collapse argument identifies: not the loss of a content that could be recovered through better instruction, but the elimination of the formative time whose irreversible quality is its constitutive property [@debord1967spectacle].
A third theoretical register makes the same argument from the side of the receiver. Benjamin’s essay “The Storyteller” (1936) distinguishes between counsel (Rat) - practical wisdom transmissible only from one experience to another, whose authority derives from the adviser’s having lived through what they speak of - and information, which arrives complete, self-explanatory, with its explanation already attached. The distinction rests on a prior one: Benjamin’s separation of Erfahrung (accumulated, reflective, transmissible experience - the practical wisdom that sediments over years of sustained engagement with consequential problems) from Erlebnis (the isolated, momentary, non-accumulating experience of the subject who encounters but does not assimilate). The formation process the tacit-as-formation register describes is precisely the process through which Erlebnis accumulates into Erfahrung: through which isolated encounters with consequential work - the model built from scratch, the case argued before a regulator, the recommendation challenged by a senior partner - sediment over time into the transmissible practical wisdom that counsel requires. The AI-mediated formation process substitutes Erlebnis without accumulation: the junior’s engagement is with AI-generated outputs rather than with the direct challenge of the work, and the Erlebnis of those engagements does not accumulate into Erfahrung - or does not accumulate at all. The apparatus of capture reduces Erfahrung to training data; what it returns - in the model’s output and in the junior’s AI-mediated engagement with it - is Erlebnis [@benjamin1936storyteller].
This distinction - tacit-as-formation rather than tacit-as-constraint - is also what rescues the Stiegler application in the consulting domain specifically: since consulting savoir-faire has always been substantially externalised into frameworks, playbooks, and deliverable templates, a content-capture reading would imply that proletarianisation was already complete long before AI arrived, which cannot be the argument. The tacit-as-formation register relocates the site of capture from content (already externalised) to the temporal structure of formative engagement - the duration of struggle, correction, and witnessed senior judgment through which the capacity to use the externalised content wisely develops - and it is this relocation that makes the Stieglerian mechanism grip a domain where content externalisation is a structural feature rather than a vulnerability.
The tacit-as-formation reading has acquired direct empirical support since the thesis was drafted from work in cognitive science. Kosmyna et al.’s EEG study of essay-writing under LLM assistance [@kosmyna2025brain] demonstrates that participants writing with AI assistance display measurably weaker brain connectivity, lower memory recall, and reduced sense of authorship over time, with the effects accumulating across repeated sessions. Oakley et al.’s review of cognitive psychology and neuroscience [@oakley2025memory] develops what the authors call the memory paradox: as external aids become more capable, the internal memory systems on which expertise, critical thinking, and long-term retention depend risk atrophy, and effective human-AI interaction itself depends on the absorbed schemata that the AI is putatively replacing. Both papers operate at the cognitive level, not the philosophical level the thesis adopts, but they provide direct empirical grounding for the structural claim the tacit-as-formation register makes: that displacing the engagements through which contributory expertise is built is not analytically equivalent to displacing the tasks themselves, because the engagements are constitutive of the formation that makes future practitioner judgment possible. Medical evidence corroborates this at the level of specific competency domains: Natali et al. (2025) identify differential diagnosis, clinical judgment, and physician-patient communication as specifically vulnerable to AI-induced upskilling inhibition - their term for the foreclosure of skill acquisition rather than the loss of already-acquired skills - precisely the formation-through-engagement capacities the tacit-as-formation account predicts will be most at risk, and their findings are explicitly extended to any professional domain where AI decision support displaces practitioner judgment [@natali2025deskilling]. Economic theory formalises why this mechanism cannot be addressed through market signals alone: Ide’s (2025) overlapping-generations model demonstrates that when entry-level work is automated, the transmission of embodied, non-verifiable knowledge between generations is structurally severed under conditions of contract incompleteness - establishing that the tacit-as-formation problem is not merely a philosophical framing but a formal economic result [@ide2025intergenerational].
The distinction also clarifies the framework’s relationship to Tuczek et al. and the wider IS-theoretical literature. The thesis is not arguing that Tuczek et al. are wrong; the analytical work they undertake at the level they operate on is productive. The thesis is arguing that the level they operate on is insufficient for engaging the question of what contemporary AI mediation is doing to European infrastructure advisory, and that the tacit-as-formation register the framework adopts is necessary for the engagement. Annex G develops this engagement, including the points of differentiation between the framework and Tuczek et al.’s Task-GenAI Fit framework.
3.6 The working ontology of the LLM
The framework treats large language models with a specific working ontology that the analytical chapters then deploy. The ontology is not a complete philosophical account of what LLMs are; it is a working specification sufficient for the analytical work the thesis undertakes.
The working ontology is this: a generic LLM is a socio-technical inscription of collective discourse, recirculated under specific conditions of ownership. Three components carry the analytical weight.
Socio-technical inscription. The model is not a technical artefact in isolation; it is an inscription of human discursive material into a technical system whose operation depends on the inscription’s content. The inscription is socio-technical in that it is technically structured but constituted by the social discursive material it encodes. The analytical implication is that the model’s outputs are not the product of the technical system alone; they are the discursive material the technical system has inscribed, recirculated through technical operation.
Collective discourse. The discursive material the model has inscribed is collective in two senses. It is collective in that it draws on human discourse at scales that no individual or institution has produced on their own - the wide corpus of internet-available text, scholarly publication, professional discourse, and adjacent material that constitutes the training distributions of contemporary models. It is also collective in that the content has been produced by communities of practice whose contributions are not individually attributable in the inscribed form. The analytical implication is that the model’s outputs are recirculations of collective work that the model itself has not produced but has inscribed and reorganised technically.
Specific conditions of ownership. The inscription operates under political-economic conditions - the institutional architectures of the firms that produce the models, the contractual architectures that govern how the inscribed material was acquired, the commercial architectures that govern how the models are deployed and monetised, and the wider political-economic context within which model production occurs. These conditions are not neutral; they shape what is inscribed, what is recirculated, and what is foreclosed. Chapter 5’s political-economic argument develops these conditions ; the working ontology here registers that the conditions are important without yet engaging them analytically.
The working ontology supports the analytical chapters in three ways. First, it grounds the framework’s tool-agnostic stance: the analytical work operates on the class of socio-technical inscriptions that contemporary generic LLMs constitute, not on any particular instance. Second, it grounds the framework’s sovereignty argument: the cultural-discursive patterns the model inscribes carry content that affects the texture of the work the model mediates, and the political-economic conditions of inscription affect what is at stake. Third, it grounds the framework’s extraction argument: the dark mirror of the SECI model operates through the inscription mechanism, where firm-distinctive material is at structural risk of being inscribed into vendor systems whose conditions of ownership are institutionally outside the firm.
The working ontology is sufficient for the analytical chapters. The deeper philosophical engagement - including the relationship between the working ontology and Floridi’s philosophy of information, the postphenomenological work on technical mediation in Verbeek and Don Ihde, what the inscription mechanism is at the level of philosophical detail - is part of what Chapter 3 bis would develop.
3.7 The framework assembled
The proletarianisation of savoir-faire through selective displacement into technical systems - Stiegler’s analytical apparatus - provides the framework’s core claim. Individuation, the process through which distinctive form is reproduced across the three scales (individual, firm, domain), provides the vocabulary for reading AI mediation as operating on formation rather than merely on output. Jonas grounds the normative commitment: answerability to futures whose conditions present decisions are shaping is the third box’s philosophical anchor.
The distinction between tacit-as-constraint and tacit-as-formation is the framework’s most direct contribution against the existing IS-theoretical literature. The existing literature asks what AI cannot substitute for; the framework asks what AI mediation does to the formation pathway through which expertise is reproduced. The questions operate at different levels. The second question is not reachable from the first.
The working ontology of the LLM as a socio-technical inscription of collective discourse, recirculated under specific conditions of ownership grounds the tool-agnostic stance, the sovereignty argument, and the extraction argument. The framework is deliberately light in philosophical register; the deeper apparatus is deferred to Chapter 3 bis and acknowledged as a limit in Chapter 9. What is established here is what the analytical chapters need. Chapter 4 describes the system the framework is now prepared to engage.
Chapter 4 takes the framework into the descriptive groundwork of how expertise is materialised in European infrastructure advisory through the integrated three-scale system. Chapter 5 takes it into the political-economic argument. Chapter 6 takes it into the comparative evidence from adjacent professions. Chapter 7 takes it into the stage-by-stage analytical core. Chapter 8 takes it into futures. Chapter 9 develops the recommendations to the four constituencies that the framework supports.