Chapter 9 — Conclusion: What to Do, and What Remains Open (plain-language version)

What this work has delivered

This closing chapter restates the claim, sets out what four groups can do about it (consultants, advisory firms, infrastructure clients and European policymakers), and says what the work has not done. The recommendations answer four of the seven findings that hold in every future Chapter 8 imagined: how advisers are formed, who controls the tools, what is owed to the public and to future generations, and the need for advisers to say openly why their work deserves trust.

The central claim and the contributions

The claim, in plain words: generic AI tools, as used today, risk hollowing out the situated, answerable judgment that makes European infrastructure advice legitimate, and on which its duties to the public and to future generations depend. The academic word for this is proletarianisation, from Stiegler, a French philosopher: practical know-how leaves the people who had it, task by task.

The chapters offer something the writing on AI in consulting lacks: one account of formation, sovereignty and the public’s stake together. It applies Stiegler’s argument to consulting at three scales; grounds the “third box”, the public and future generations to whom advisers must answer, in the ethics of Jonas, a philosopher of responsibility; maps the profession’s competency standard (ChMC) against the work AI can replace; shows “pipeline rupture”, the loss of the junior work that forms tomorrow’s seniors, across professions; and offers four scenarios as thought-experiments.

One trap must be named first. Debord, a French thinker, described how a critique gets neutralised by being turned into a show of response 1 . In advisory work the show is familiar: a responsible-AI policy, a junior who checks the model’s output, a senior who signs it off. None of it gives back the time spent struggling, unassisted, with a hard problem, and that time is how judgment forms. Blangeois and Lebrument read the coherent AI story that firms billing human effort tell on their earnings calls as “a rhetorical performance” 2 💡 . The recommendations try to guard against it.

What this means for consultants

In general form, these recommendations hold in any advisory field where AI is absorbing the work that used to form juniors; European infrastructure, where the public’s stake is heaviest, makes them most demanding. One example: a junior consultant accompanies a senior partner to a European Commission working group, to watch how he handles a question nobody prepared for. In the taxi afterwards they spend forty minutes on it. Years later she calls it the most formative afternoon of her career, and she is certain no AI tool could have given it to her 💡 .

The first recommendation is deliberate formation. Juniors should seek work that builds real expertise through contact with consequences, even when a tool would be faster. Mid-career consultants form the juniors, which asks more than reviewing machine-assisted drafts. Seniors should mind the tasks they refuse to hand to a tool. The second is reflexive practice: noticing how the tool changes your analysis and your own grasp of the problem, without refusing AI. The third is to refuse unattributed AI work. The junior owes attribution to the reviewing team, the mid-career consultant owes it to the client, and the senior owes it to the public who bear the consequences. The fourth is to cultivate judgment that resists being averaged out: the ways of seeing that are yours, and that generic tools flatten. In practice, know which parts of the work your voice comes from, and keep doing those parts yourself. The fifth, mostly for seniors, is responsibility for the next generation: invest in formation, take supervision seriously, and recognise that the firm’s capacity in twenty years is being decided now. Juniors learn what a senior is by watching one: a partner who checks the model’s draft, signs and forwards it is teaching that this is the job.

What this means for advisory firms

The same logic holds for any advisory firm whose pipeline of future seniors is wearing away at its base. The first recommendation is to preserve formation pathways on purpose: decide which projects keep juniors in real contact with the work, and which supervision they get. The second is to protect the firm’s knowledge from extraction. A practice choosing an AI platform can make it a condition that no client content goes into the vendor’s training data. The third is to cultivate the firm’s distinctive voice as an asset: its way of seeing, its methods and its instincts for the sector. That voice lives in the people the firm forms: a style guide or a prompt library can record it, but cannot renew it once formation stops.

The fourth is explicit attribution at firm level: telling clients and colleagues how the work was produced. Mizrak and colleagues found that people hide AI use for fear of looking less competent 3 , and the measure they rank first is a disclosure system that does not punish 3 . 💡 The fifth is to take the formation question to industry level, with chartering bodies and regulators. The trap applies here too: the aim is investment in formation itself, not more machinery of oversight. 💡

What this means for infrastructure clients

Infrastructure clients carry public-interest duties and long time horizons, and they answer to the third box in a way more commercial clients do not. The first recommendation is to procure for judgment, not only for output. AI produces fluent talk far more readily than real expertise, so buying rules should tell the two apart. Benjamin, a German critic, distinguished counsel, the practical wisdom that passes only from one person’s long experience to another’s, from information, which any model can produce 4 . The client is buying counsel: work with an “aura” 💡 , the mark of real formation 5 .

The second is disclosure that recognises the third box. The public who bear the consequences of infrastructure decisions are heard through consultations and regulatory submissions, and those only function if the use of AI is made visible. The third is to think strategically about in-house AI capability, because it reshapes the advisory world clients depend on. The fourth is to audit that capability’s effect on formation. If clients keep only the high-judgment tasks for outside firms, nothing is left that forms juniors, and the client builds on eroding ground 💡 .

What this means for European policymakers

The general point: treat the formation of advisers who serve the public interest as a policy concern, not a private matter for firms. The first recommendation is to learn from defence, where European policymakers already treat sovereignty as strategic, and to extend that logic, adapted, to other infrastructure fields. The second is to build sovereignty into procurement. Mügge shows that EU strategy has treated independence from foreign jurisdictions as a stand-in for control over the knowledge inside the tools 6 . Procurement is never neutral: it should recognise the public-interest tradition, the third box and the formation pathway.

The third is to take up formation at policy level, through support for chartering and accreditation bodies and through education. The European tradition lasts only as long as people are formed inside it, so formation policy is also sovereignty policy. This is more diffuse, but it matters. 💡 The fourth is to use existing instruments: the AI Act’s rules for high-risk systems in critical infrastructure 7 and the Green Deal’s frameworks for energy, transport and water could carry formation requirements. The fifth is to give the third box explicit institutional standing. The public and the generations to come are the people advisers finally answer to; policy can give them a place through consultation requirements and the explicit recognition of responsibility between generations.

The limits of this work

Three limits mark where the contribution stops. First, it covers a single domain: the evidence from software and law suggests the mechanism reaches further, but the framework is built for European infrastructure advisory. Second, the work is reflexive and conceptual, not systematically empirical: no interview programme, no survey, no long-term study. Third, the philosophy is deliberately light. It is enough for the analysis done here, but it is not the complete framework.

The questions left open

Six questions remain. The first, and the most consequential, is to test the predictions through interviews and long-term studies. The second is comparison with other kinds of consulting and with non-European settings. The third is the deeper philosophy. The fourth is the operational detail of the recommendations. The fifth is the long-horizon claim that today’s damage to formation will show in the senior advisers of two decades from now, and in how they come to see themselves, a question Ke and colleagues leave open for medicine 8 . No single research programme can check it, which is not a failure: responsibility to the future does not wait for proof, and the claim about 2046 is a commitment, not a forecast. 💡

The sixth is whether paying advisers by results is compatible with answering to the third box. The profession’s own foresight expects results-priced advice to become the template for other services 9 . But a contracted result is one two parties have agreed to measure, and the public and future generations are party to no contract. A fifty-year asset does not clear in a results clause.

A closing word

I write as a consultant. The itch that drove this work is one I am experiencing. The framework is offered as a way of seeing a change that productivity talk cannot capture. I do not know which of the four futures will arrive.

The concerns named here do not belong to one school of thought. Leo XIV’s 2026 encyclical on AI reaches a related formulation from within Catholic social teaching 10 . That is some reason to think the concern is more than an artefact of my method, though traditions can share a blind spot as readily as an insight. The image from the film WALL-E 11 💡 names the far end of the mechanism: a profession, and beyond it a society, that has given up its capacity for situated judgment and can no longer exercise it unaided.

I hope this work can be one piece of a wider conversation already under way. The direction is being set now, by the decisions practitioners and institutions are making. Engaging seriously with the stakes is the European public-interest tradition itself, and the tradition has been kept alive that way for decades. Whether it continues is not a question this work can close.

The work that follows will answer it… hopefully! 💡