Glossary (plain-language version)
A guide to the key concepts used in this thesis, written for readers without a philosophy or academic background. Terms are grouped by theme.
I. The core argument
Proletarianisation (of advisory)
The gradual transfer of practical expertise, judgment, and professional identity out of the people doing a job and into technical systems they donât own. The word comes from the philosopher Bernard Stiegler, who borrowed and reworked it from Marx. It doesnât mean job losses here. A factory worker was âproletarianisedâ when the craft knowledge in their hands was encoded into machines they operated but didnât design or control. The thesis argues something similar is happening to consulting: not that consultants are being replaced, but that the specific activities through which they have historically developed their judgment are being absorbed into AI tools. The expertise doesnât disappear â it just stops living in the people.
Selective displacement
How proletarianisation actually happens in knowledge work: not by replacing whole jobs but by taking over specific components of the work. AI tools do routine analysis, draft documents, summarise research â while leaving senior judgment and relationship work to humans. The important thing is which components get taken: the ones AI tools absorb tend to be exactly the ones that used to form junior practitioners into experienced ones.
Temporal structure collapse
The specific form the problem takes in consulting. Unlike a factory, consulting has always had a lot of its knowledge written down â in frameworks, playbooks, training materials. So AI doesnât capture consulting âsecretsâ so much as it collapses the time it takes to do the work. A financial model that used to take a junior consultant three weeks to build â weeks of struggling, making mistakes, asking questions, getting corrected â now takes hours. The output appears. But the formation that used to happen during those three weeks doesnât.
II. The philosophical framework
Tertiary retention
Stieglerâs term for external memory objects â books, recordings, databases â that exist outside a personâs mind but shape how they think. The danger he identifies is when these memory objects are controlled by someone else and start substituting for experience rather than supplementing it. In the consulting context: when a junior consultant uses an AI-generated analysis instead of working through the problem themselves, they receive the output but skip the experience that historically produced it.
Individuation / dis-individuation
The philosopher Gilbert Simondon described becoming a person â or a professional â not as arriving pre-formed but as an ongoing process of developing through engagement with other people and with the world. A junior consultant becomes a senior consultant through years of consequential work, challenge, mentorship, and error. Stiegler called the impoverishment of this process âdis-individuationâ: when technical systems reduce a personâs role to operating a tool rather than genuinely engaging with a problem, they stop developing in the way they otherwise would. The thesis uses this idea at three levels: individual practitioners, consulting firms (which also have an identity that can be diluted), and the wider European professional tradition.
Identity (three scales)
What individuation has produced so far, at each of the three levels: the consultant as a particular kind of practitioner, the firmâs own way of seeing, and the European public-interest tradition. It is provisional, because the same process can still change it, and the thesis does not defend it for its own sake. The idea that a professional identity forms over time, through experience and feedback, comes from the organisational researcher Herminia Ibarra. AI tools act on the process and so on its result: a junior may come to see the job as passing on what the model produced, a firmâs voice may drift toward the modelâs defaults, and the tradition may be pulled toward ways of framing problems formed elsewhere. The trial versions of the role that juniors try on may come from the model rather than from the seniors they watch.
Savoir-faire / savoir-vivre / savoir-penser
French terms Stiegler uses for three kinds of know-how at risk of displacement. Savoir-faire: practical skill â how to do the work. Savoir-vivre: how to conduct oneself â the ethical and cultural dimensions of professional life. Savoir-penser: how to think â the capacity for independent analysis rather than just operating a system. The thesis argues AI mediation in consulting threatens all three, not just the first.
Symbolic misery / noetic proletarianisation
Stieglerâs broader terms for what happens when the capacity to produce meaning and to think independently gets hollowed out by technical systems. The thesis situates its consulting argument within this wider concern before zooming into the specific mechanism that applies to consulting.
III. Ethics and moral responsibility
Third box
The thesisâs own term for a constituency that doesnât usually have a seat at the table in advisory work: the communities, future generations, and affected publics who bear the consequences of infrastructure decisions without being party to the contracts.
In most consulting engagements there are two parties: the client (who pays) and the consultant (who advises). But in infrastructure â a new railway, a water tariff reform, an energy grid â there is a third group: the people who will live with the result for decades. They didnât hire anyone. They may not have been consulted. But their lives are shaped by decisions made on their behalf. The thesis argues that legitimate infrastructure advisory has always carried responsibility to this third group, even when itâs not explicitly named, and that AI mediation makes this responsibility harder to honour.
Jonasian responsibility
Hans Jonas was a philosopher who argued that modern technology has created a new kind of moral situation. When our actions can affect people far in the future â people not yet born, who have no voice in current decisions â we have obligations to them that older moral frameworks didnât address. The more power you have to shape the future, the more responsibility you carry for the conditions you create. In the thesis, this grounds the argument that infrastructure advisers carry responsibilities not just to their clients but to the future generations who will live with the consequences of what they recommend today.
Internal goods / practice (MacIntyre)
The philosopher Alasdair MacIntyre distinguished between two kinds of things people get from their work. External goods â money, status, reputation â can be obtained by any means and donât depend on the quality of the work itself. Internal goods are different: they are the forms of excellence that can only be achieved by genuinely engaging with the work on its own terms. A consultant who shortcuts their analysis to produce a polished deck faster captures external goods (a satisfied client, a fee) but loses the internal goods â the satisfaction of genuinely understanding the problem, the development of judgment that only comes from wrestling with it. The thesis uses this to argue that what AI mediation most directly threatens in consulting is the internal goods: the things that make the profession worth doing and make its outputs worth trusting.
IV. Knowledge and expertise
Tacit knowledge
Knowledge we have but canât fully put into words. The philosopher Michael Polanyiâs famous phrase: âwe know more than we can tell.â A skilled consultant knows when a clientâs briefing is missing something important, or when a regulatory bodyâs question signals a concern they havenât stated. This knowledge comes from experience and canât easily be written down as a rule. The question the thesis engages is what happens to this kind of knowledge when AI tools take over many of the activities through which it is normally developed.
Tacit knowledge as constraint vs. tacit knowledge as formation
Two very different ways of understanding tacit knowledge, with opposite implications.
The âconstraintâ view: tacit knowledge is just information thatâs currently hard to transfer, but better technology can eventually capture it. Under this view, AI tools that make tacit knowledge more explicit are simply useful.
The âformationâ view: tacit knowledge isnât information at all â itâs the result of a process. The knowledge doesnât exist as content that could be extracted; it exists as a capacity that develops through sustained engagement with difficult work over time. Under this view, bypassing the formative process doesnât just skip a step â it eliminates what made the step valuable.
The thesis argues the formation view is correct for consulting, and that this is the reading the existing research has mostly missed.
Contributory expertise
A distinction developed by the sociologist Harry Collins. Interactional expertise means knowing enough about a field to communicate competently with its practitioners â you can speak the language, ask the right questions, follow the argument. Contributory expertise means knowing enough to actually contribute to the fieldâs development â to add something genuinely new, to make independent judgments, to push the thinking forward. The thesis argues that what AI tools have become very good at is interactional expertise: producing outputs that sound like consulting. What they cannot do is contributory expertise â the kind of judgment a senior consultant develops through years of engagement with genuinely consequential work.
Counsel vs. information (Walter Benjamin)
The cultural critic Walter Benjamin distinguished two modes of transmission. Counsel is practical wisdom embedded in the adviserâs lived experience â it comes from somewhere, and its meaning is completed by the receiverâs own experience in engaging with it. Information arrives complete and self-explanatory; it doesnât require experience on either side to produce or receive. Benjamin was writing about storytelling and journalism, but the thesis applies the distinction to consulting: what a formed senior adviser gives a client is counsel â judgment rooted in experience, carrying the trace of that experience. What AI-generated advisory output delivers is information. Sophisticated clients can often sense the difference even if they canât articulate it.
Erfahrung and Erlebnis
Two German words Benjamin used for two different kinds of experience. Erfahrung is accumulated, reflective experience â the practical wisdom that builds up over years of sustained engagement and makes counsel possible. Erlebnis is momentary, isolated experience that doesnât accumulate â the experience of encountering something without it becoming part of you. The thesis uses this pair to describe what AI mediation does to junior consultants: they encounter many problems and produce many outputs, but the engagement is with AI-generated material rather than with the direct challenge of the work, and the experience doesnât accumulate into the kind of judgment that took decades to develop in previous generations. You get Erlebnis without the Erfahrung it should be building toward.
V. Professional formation
ChMC (Chartered Management Consultant)
The professional qualification framework for management consultants that the thesis uses as its analytical scaffold. It describes three stages:
- Foundation: Junior practitioners learning the basics through directed analytical work â building models, structuring outputs, executing defined tasks under supervision.
- Applied: Mid-level practitioners managing delivery â translating between senior judgment and junior execution, developing the contextual and relational intelligence to make analysis useful in real situations.
- Chartered: Senior practitioners exercising authoritative judgment â contextually trusted, grounded in what they owe the public, capable of disagreeing with the client when necessary.
The thesis traces what AI mediation does at each stage.
Pipeline rupture
The thesisâs term for how problems at the junior stage propagate to the senior stage over time. If Foundation-stage consultants are no longer doing the formative work that historically developed their judgment â because AI tools are doing it instead â then the cohort that should be developing into experienced senior consultants over the next decade is not developing in the way it should. Current senior consultants are fine; they were already formed. But the pipeline that should be producing the next generation of senior consultants is disrupted. This is invisible in short-term measures of output quality, but it is structurally consequential. Ideâs economic model gives a formal account of how automating entry-level work can weaken the passing of unwritten know-how between generations, even while entry-level jobs are kept.
Upskilling inhibition
The phenomenon where AI assistance, while improving the immediate quality of output, inhibits the development of the underlying competence that unassisted work would have built. Research in medical training has demonstrated this: junior doctors whose assessments are AI-assisted produce better-looking results but develop less clinical judgment than those who wrestle with assessments themselves. The thesis applies this to consulting: fluent output produced with AI assistance may mask the absence of the formation that would historically have accompanied producing it.
Cognitive debt
Research in neuroscience and cognitive science has found that AI-assisted thinking involves less neural engagement than unassisted thinking â and that this difference persists even after AI assistance is withdrawn. In other words, using AI tools to do cognitive work doesnât just save time; it may change how deeply the work is processed and retained. The thesis uses this as part of the evidence that engagement with AI-mediated work is structurally different from engagement with unassisted work.
VI. Organisations and political economy
Sovereignty (advisory)
The thesis uses this word in a specific sense: the capacity of an adviser to maintain an independent interpretive standpoint â to disagree with the clientâs framing, to represent the interests of people not in the room, to exercise judgment that isnât reducible to whatever the AI toolâs default outputs happen to be. This operates at three levels: the individual practitioner (who can say âI think youâre wrongâ), the firm (which has its own distinctive way of understanding problems, different from other firmsâ), and the wider European professional tradition (which has its own values, assumptions, and ways of weighing public interest, distinct from the Anglo-American patterns dominant in most AI training data).
The firmâs way of seeing
Each consulting firm has developed, over decades of work, a distinctive interpretive framework â a characteristic way of analysing situations, framing problems, calibrating recommendations. This isnât just a brand or a style; itâs embedded in the people, the practices, and the culture of the firm. The thesis argues that this distinctiveness is at risk when firms routinely use generic AI tools as their analytical starting point, because those tools carry their own defaults and framings â and repeated use gradually pulls the firmâs outputs toward those defaults.
SECI model / dark mirror
The SECI model (from management researchers Nonaka and Takeuchi) describes how organisations create and transmit knowledge: tacit knowledge gets made explicit, combined with other explicit knowledge, and then re-internalised by practitioners as new tacit knowledge â a virtuous cycle. The thesis proposes a âdark mirrorâ of this cycle: when firms externalise their knowledge into AI tools (through prompts, outputs, shared documents), that knowledge can flow back into the training data of AI systems owned by others. The virtuous knowledge cycle becomes an extraction mechanism. What the firm knew becomes what the AI tool knows â and the firm has transferred its distinctive capability to an external system it doesnât control.
Aura (Benjamin)
Benjamin used âauraâ to describe the quality of authentic presence that derives from something being embedded in a specific tradition â its âhere and now,â the traces of its particular history. A senior consultantâs advice carries aura in this sense: it comes from a specific personâs sustained engagement with a specific tradition, a specific set of clients, a specific regulatory environment. AI-generated advice lacks this aura: it draws on the generic patterns of a training corpus rather than the traces of a particular tradition. Benjaminâs own analysis is ambivalent â mechanical reproduction democratises, making things available that were previously scarce. The thesis engages this ambivalence: AI tools do make competent analytical output more widely available (a real gain), but they cannot reproduce auratic counsel (a real loss). Both are true simultaneously.
Apparatus of capture (Deleuze & Guattari)
Two philosophers, Gilles Deleuze and FĂ©lix Guattari, described how large institutional structures â states, corporations â absorb, redirect, and monetise flows of activity that would otherwise remain independent. The thesis uses this concept to describe what is happening to professional knowledge under AI mediation: the practical know-how, judgment, and formative engagement of consultants doesnât just disappear; it gets redirected into proprietary AI architectures owned by actors outside the profession and outside European public-interest governance. The capture is systematic and structural, not incidental.
Spectacular time / récupération (Debord)
The cultural critic Guy Debord, writing in the 1960s, described how capitalist culture colonises time â specifically how the rhythm of product releases and novelty cycles prevents the kind of slow, accumulated experience that builds real understanding. The thesis uses this to name something familiar to anyone who works with AI tools: the constant rhythm of capability releases, model updates, and new features creates a sense of perpetual renewal that structurally forecloses sustained, non-spectacular engagement. Each new model appears as progress; the practitioner is perpetually adapting rather than deepening.
RĂ©cupĂ©ration is Debordâs term for how radical or critical ideas get absorbed by the dominant culture and turned into commodities. The formation argument, the sovereignty argument, the recommendation to preserve junior practitionersâ formative engagement â all of these risk being rĂ©cupĂ©rĂ©: praised by firms, packaged into training modules, turned into a marketing claim, and thereby emptied of the challenge they pose. The thesis flags this as a genuine risk for any institutional response to the argument it makes.
TGAIF (Task-GenAI Fit Framework)
A research framework developed by Tuczek and colleagues (2026) for assessing which tasks are well-matched to AI assistance, by sorting them on two dimensions: how uniform or flexible a task is, and how general or specific. The result is four task profiles, such as the âScalable Generalistâ and the âAdaptive Specialistâ. The thesis treats this as the main example of the task-focused approach to AI in consulting. It reads unspoken, tacit knowledge as a limit on what can be automated, and recommends how much to automate and how much to assist for each profile, allowing deliberate non-use for sensitive tasks. It is useful for matching tools to tasks, and its authors do mention the risk to how juniors build skills, but their recommendations leave it aside, so what happens to junior practitioners when the tasks that used to form them go to AI tools stays out of view.
VII. Context
European infrastructure advisory
The specific domain this thesis studies: advisory work on European infrastructure â railways, energy systems, water, urban development, defence, environmental projects. Defined by three features that make the stakes particularly clear: (1) a tradition of working in the public interest, rooted in post-war European reconstruction; (2) regulatory frameworks oriented to public outcomes rather than purely commercial ones; (3) significant âthird boxâ exposure â the decisions made here affect communities and futures in ways that management consulting to a retailer, for example, typically doesnât.
Pyramid collapse
The traditional consulting firm employs many junior consultants (billed at lower rates, doing analytical work) to support fewer senior ones (billed at higher rates, exercising judgment). This pyramid structure was financially efficient â and it was also where junior formation happened. AI tools are collapsing this pyramid from below: as junior analytical work gets absorbed by AI tools, firms need fewer junior consultants. This changes the firmâs economics â and it also changes the formation pathway. The pyramid collapse is simultaneously an economic shift and a formation crisis.
Symbolic domination
The thesisâs term for a specific cultural risk: when AI tools built primarily on Anglo-American data and assumptions are used routinely by European consultants, the result is more than an efficiency gain. The tools carry particular ways of framing problems, particular rhetorical conventions, particular assumptions about what a good recommendation looks like. Over time, routine use of these tools can gradually pull European advisory practice toward the cultural defaults embedded in the tools â displacing the distinctively European way of weighing public interest, regulatory complexity, and long-term systemic risk that has historically distinguished European infrastructure advisory.