Annex A — Proletarianisation
A.1 Why this annex exists
Proletarianisation is the idea this thesis leans on hardest. I use it as reworked by Bernard Stiegler, a French philosopher 1 Stiegler (2016) Automatic Society, Volume 1: The Future of Work. Polity Press. : practical know-how displaced into systems the practitioner does not own.
A.2 Where the word comes from, and why Marx’s version is not enough
For Marx, the word meant workers losing control of their tools. A consulting partnership is not a factory. A later idea of Marx’s does reach expert work under AI 2 Ngwenyama et al. (2026) Platform capture of scientific knowledge production: publishers’ dominance, generative AI and subsumption of academic labor. European Journal of Information Systems. 35(5), 942–965. p. 945. , but it asks who owns the knowledge. Stiegler asks what happens to the knower. 💡 An examiner who knows the Grundrisse will ask this in the first ten minutes. Better to have answered it in the annex than to improvise it in the viva.
A.3 Stiegler’s reworking
For Stiegler, the loss is of know-how itself, captured into systems owned elsewhere. It reaches three kinds 1 : knowing how to do the work, how to live, and how to think.
A.4 Selective displacement: the qualification this work treats as central
Only parts of the work move into the tools: routine analysis and the formative tasks of early careers. Situated judgment stays. Nothing looks wrong from outside; the change shows in how people are formed, not in headcount. Predictions of wholesale substitution 3 Susskind et al. (2015) The Future of the Professions: How Technology Will Transform the Work of Human Experts. Oxford University Press. miss it.
A.5 How this work uses the term
In European infrastructure advisory, it means hands-on expertise selectively displaced into AI systems owned elsewhere, with consequences for formation and sovereignty.
A.6 What this work is not claiming
I do not claim that AI will abolish consulting, that it should be refused, or that the outcome is fixed. Consultants, firms, clients and policymakers still choose.
Annex B — Third Box Thinking
B.1 Why this annex exists
The third box is my name for the people who live with infrastructure decisions without holding the contracts behind them: the public, and future generations whose lives are being decided now. Advisers answer to them.
B.2 The structural problem the concept addresses
An engagement has two visible parties: the firm and the paying client. A water tariff reform shapes household budgets, yet those households signed nothing. 💡 From inside practice, the gap between acknowledging this asymmetry and actually doing something about it is wider than the concept implies. I have sat in project inception meetings where the client’s brief was the only frame in the room. The publics who would live with the outcome were present as a regulatory compliance requirement, not as a genuine constituency. The third-box concept is partly a diagnostic for why that happens so routinely — not just a normative aspiration for what should happen instead.
B.3 The Deweyan ground: how publics come into being
John Dewey, an American philosopher, argued that a public forms when people share exposure to the consequences of others’ decisions 4 Dewey (1954) The Public and Its Problems. Alan Swallow. . So the third box is already there.
B.4 The Jonasian register: responsibility toward the future
Hans Jonas, a philosopher, extended responsibility to everyone whose life a present decision will shape, including future generations with no voice today 5 Jonas (1984) The Imperative of Responsibility: In Search of an Ethics for the Technological Age. University of Chicago Press. . A fifty-year asset decided in a five-year political cycle binds them.
B.5 Why the three-box framing matters
With it, advisory work can be judged on how it answers to these people. Who they are changes with the project; the position does not.
B.6 Third-box thinking as a habit of analysis
The third box is people. Third-box thinking is the habit of asking, at every step, what the analysis means for them.
B.7 How this work uses the concept
The idea returns throughout the chapters, because what advisers owe the public and future generations is where the argument always comes back.
B.8 What the concept does not claim
The third box has no single identity, and a consultation run only to satisfy a procurement condition does not reach it. The idea works within Europe’s democratic institutions; the worry is that AI makes these people harder to reach.
Annex C — The ChMC Framework
C.1 Why this annex exists
ChMC stands for Chartered Management Consultant, the UK framework for chartered status in consulting, launched in 2021 by the Chartered Management Institute 6 Chartered Management Institute (2024) Chartered Management Consultant (ChMC) Competency Framework. . It anchors this thesis’s definition of consulting, and its three career stages organise Chapter 7.
C.2 What the ChMC framework is
The framework defines consulting, sets out the competence expected at each career stage, and is the route to chartered status 6 . It crosses four areas of competence with three stages.
C.3 The four competency components
The four areas are ethics and professional standards, leadership and management, understanding the client’s environment, and personal and professional development. Ethics is where answerability to the third box comes in: the public and future generations who live with the decisions. Development is the area AI disrupts most, because it is the formation system itself.
C.4 The three progression stages
Foundation, roughly the first three to five years, is formative work under supervision. Applied, roughly years five to twelve, is running engagements and supervising juniors. Chartered, the second decade and beyond, is where judgment is the main contribution. Each stage builds on the one before. The year ranges fit UK careers better than continental ones.
C.5 The framework in comparative context
💡 This is where I feel the limits of the ChMC as an organising scaffold most acutely. The three-stage Foundation/Applied/Chartered architecture is legible and clean, and it works as analytical scaffolding. But it was designed with UK consulting in mind, and the career structures in continental European infrastructure advisory — where entry routes run through grandes écoles, through public-sector planning bodies, through Rijkswaterstaat or equivalent national institutions — do not map onto it tidily. The ChMC stages are a useful analytical frame for this thesis. They are not a faithful description of how careers actually unfold in the contexts I know best. That gap is worth sitting with rather than papering over. Engineering chartership, the French elite schools, and Dutch and Nordic public institutions each form practitioners differently. I use the ChMC because it was written for consulting. A newer scaffold, a HEC Alumni white paper, proposes a twelve-month pathway for the AI-assisted junior consultant 7 Club Consulting & Coaching (C3), HEC Alumni (2026) ConseilIA : le nouvel âge du Conseil Augmenté. Réinventer le Conseil : comment l’IA redessine le métier, les compétences et l’avenir du secteur. pp. 209–211. . Beside a Foundation stage of several years of supervised client work, it is competent training in checking machine output. It is not formation.
C.6 Limits of the framework as an analytical scaffold
The framework is young, general, and anchored in the UK. Carrying it to Germany, France or the Mediterranean countries is translation I have not done. It is enough for what this thesis asks of it.
Annex D — Individuation
D.1 Why this annex exists
Individuation is the process by which a practitioner, a firm or a professional tradition takes a form of its own and keeps it. It never finishes. I use it at three levels, to read AI as one mechanism at work in three places.
D.2 Where the concept comes from: Simondon
The idea comes from Gilbert Simondon, a French philosopher 8 Simondon (2020) Individuation in Light of Notions of Form and Information. University of Minnesota Press. : the individual is the product of this process, not its starting point. Form emerges from a field that is stable enough to hold yet ready to reorganise, a condition he calls metastable. 💡 Metastability is, for my purposes, the most generative concept Simondon offers. A system that is merely stable is closed — it has resolved and has nothing left to become. A system in metastable equilibrium is stable enough to function and sustain itself, but genuinely open to transformation when new information or energy enters the field. This is precisely what I need the individuation framework to do: it must describe entities (practitioners, firms, traditions) that are formed and coherent, not blank slates, but also genuinely capable of being altered by the technical mediations they encounter. Pure process concepts cannot account for the stability; pure identity concepts cannot account for the openness. Metastability holds both.
D.3 Stiegler’s extension
Stiegler added that we take form through the technical systems we work with 9 Stiegler (1998) Technics and Time, 1: The Fault of Epimetheus. Stanford University Press. . When know-how is captured into systems owned elsewhere, that process is interrupted 1 .
D.4 How this work uses the concept
A consultant becomes a practitioner through formative work. A firm becomes itself by passing on its way of seeing. The European public-interest tradition keeps itself alive through its institutions. The result so far is what the chapters call identity. Each is still forming, so generic AI can deform all three.
D.5 What the concept does not claim
The process is not good in itself, and nor is its result, identity: the argument does not defend the profession as it now is. Technology is not its enemy, and the damage is selective, not total. The concern is who owns the tools.
D.6 Where the framework is incomplete
Simondon’s deeper philosophy, his neighbours, and how the three levels join are left for later work.
Annex E — Tacit and Contributory Expertise
E.1 Why this annex exists
This annex sets out the two sources behind the thesis’s most direct contribution: two ways of reading tacit knowledge, the knowledge that cannot be fully put into words.
E.2 Polanyi and the tacit dimension
Michael Polanyi, a philosopher, gave the canonical phrase: we know more than we can tell 10 Polanyi (2009) The Tacit Dimension. University of Chicago Press. . A profession’s knowledge includes judgment that cannot be written down, so writing things down does not replace the practitioner.
E.3 Collins on contributory and interactional expertise
Harry Collins, a sociologist of science, separates being able to do the work from being able to talk about it fluently 11 Collins et al. (2007) Rethinking Expertise. University of Chicago Press. . AI produces the fluency fast. It does not produce the doing. 💡 In consulting practice, the difference between contributory and interactional expertise is immediately recognizable to anyone who has worked in a team over time. The person with interactional expertise knows the language of the domain, can follow the reasoning, and can hold a credible conversation with specialists. The person with contributory expertise can actually produce the analysis — they know when the numbers are wrong before the model has finished running, they recognize the pattern in the client’s hesitation, they know which regulatory risk is real and which is a formality. I have sat in rooms with both kinds of people. The gap is not subtle. What concerns me about AI-mediated formation is that it accelerates the production of interactional fluency while leaving contributory competence underdeveloped.
E.4 This work’s distinction: tacit knowledge as constraint versus tacit knowledge as formation
Most research on information systems treats unspoken knowledge as a limit on what AI can take over 12 Tuczek et al. (2026) Where Automation Meets Augmentation: Balancing the Double-Edged Role of Generative AI in Management Consulting. Business & Information Systems Engineering. .
I read it as the substance of how a consultant becomes one, and ask what handing formative tasks to AI does to that.
E.5 How this work uses the concepts
Chapter 3 builds the distinction. Chapters 4 to 9 apply it, from junior formation to the recommendations.
E.6 What the concepts do not claim
Not all professional knowledge is unspoken, fluent talk has its uses, and none of this is mystical: it is built through years of consequential work.
Annex F — The SECI Model
F.1 Why this annex exists
The SECI model, named after its four steps (socialisation, externalisation, combination, internalisation), describes how a firm passes its knowledge from one generation of practitioners to the next. Chapter 4 uses it. Chapter 5 turns it over to show how that knowledge can flow out of the firm.
F.2 What the SECI model is
Nonaka and Takeuchi, writing about Japanese firms, named four steps 13 Nonaka et al. (1995) The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation. Oxford University Press. : learning beside a senior, writing know-how down, recombining written material in new work, and absorbing it through use. Together they form a loop 14 Nonaka (1994) A Dynamic Theory of Organizational Knowledge Creation. Organization Science. 5(1), 14–37. .
F.3 The model’s application to consulting
Through that loop a firm reproduces itself: its methods, its memory, its voice. Disrupt the loop and you disrupt the firm’s identity, not just its output.
F.4 The dark mirror: how SECI becomes extraction
The model assumes written-down knowledge stays inside the firm. When drafts and frameworks pass through an outside vendor’s AI tools, that depends on the vendor’s terms. This is the dark mirror: the firm’s knowledge flowing outward. It is an argument about where the pathway lies, not a record of documented leaks. 💡 This is the most practically actionable concern in the thesis. Unlike the formation argument, which plays out over decades, the extraction pathway through vendor terms of service is immediate and addressable now. I have had direct conversations with partners about exactly this: the moment they realised that the working draft sent through a particular tool might be feeding a model that their competitors also use. The reaction is not abstract concern — it is recognition. The structure was already there; they simply had not named it.
F.5 How this work uses the model
Chapters 4, 5 and 7 use the model to read what AI does to a firm’s identity.
F.6 Limits of the model for this work’s purposes
The model was built for Japanese firms, ignores who owns what, and says little about what passes through the loop. That reading is mine.
Annex G — TGAIF and the Tuczek et al. Framework
G.1 Why this annex exists
TGAIF, short for Task-GenAI Fit, is a 2026 framework by Tuczek and colleagues that sorts consulting tasks by how far generative AI can take them over 12 . It is the closest work to this thesis, and I treat it as serious work to run alongside.
G.2 What the TGAIF framework is
It sorts tasks two ways: done the same way every time or worked out afresh, and generic or needing real knowledge of the field. Uniform, generic tasks go to the tool. Flexible, specialist ones stay with the person.
G.3 What the paper does and does not do
The paper is concrete and grounded in evidence, and it uses Polanyi’s idea of unspoken knowledge with care 10 . It notes that automating junior roles “challenges traditional pathways for skill development” 12 Tuczek et al. (2026), p. 8. , but its recommendations leave that aside. The framework has no place for formation, for who owns the AI, or for sovereignty. These are choices, not failures.
G.4 This work’s TGAIF overlay
Laid over the career stages, junior work falls where AI fits best. That is exactly where I locate the damage to apprenticeship. Both are true. 💡 The structural irony here is the most consequential point of divergence between the two frameworks. TGAIF categorises Foundation-stage tasks as high AI fit — correctly, on its own analytical terms. This thesis argues that the automation of precisely those tasks is the mechanism of formation rupture. The efficiency that TGAIF identifies at the Foundation stage is the same dynamic that this thesis identifies as the apprenticeship blow. The frameworks are not disagreeing about the empirical situation; they are disagreeing about what question to ask about it.
Senior work is where the two analyses agree: it is judgment a tool cannot supply.
G.5 What the differentiation amounts to
TGAIF asks which tasks AI can do and counts efficiency as success. I ask what happens to formation, and count as success advisory work that still answers to the third box: the public and future generations who bear its consequences.
G.6 Where the engagement could be developed further
A wider comparison with research on information systems 15 Tronnier et al. (2025) Consulting in the Age of AI: A Qualitative Study on the Impact of Generative AI on Management Consultancy Services. Proceedings of the 58th Hawaii International Conference on System Sciences. and use of TGAIF’s evidence remain future work.