There are moments when the future arrives not as a prophecy, nor as a cinematic catastrophe, but as an executive classification scheme. Builders build. Sellers sell. Measurers measure. The formula is neat, memorable and slightly terrifying, which is often how history chooses to introduce a new social order. Matthew Prince, the CEO of Cloudflare, recently explained why his company had laid off more than a fifth of its workforce despite strong growth and healthy revenues. It was not, he argued, because Cloudflare was failing. It was because artificial intelligence had changed the shape of the company itself. Some people were still needed to create products. Some were still needed to sell them. But the large category of people who measured, coordinated, checked, reported, audited, managed and interpreted the organisation to itself could now, in many cases, be replaced or compressed by AI systems.
At almost the same moment, Pope Leo XIV issued Magnifica Humanitas, an encyclical on artificial intelligence and the safeguarding of the human person. The contrast was too perfect to be ignored. On one side stood the CEO, looking at the firm and asking how it could be made leaner, faster and more competitive. On the other stood the pope, looking at the human person and asking what must never be treated as disposable. One was speaking the language of operating models. The other was speaking the language of dignity. They are not discussing different technologies. They are looking at the same transformation from opposite ends of the moral telescope.
The tempting response is to turn this into a simple morality play: Prince as the cold capitalist and Leo as the defender of humanity. That would be satisfying, and therefore probably too easy. Prince is not wrong that many organisations have become thick with measurement. Over the last half-century, large companies, universities, hospitals and public administrations have accumulated layer upon layer of managers, analysts, compliance officers, coordinators, risk teams, performance frameworks, reporting processes and internal communications structures. Some of this work is essential. Some of it is the inevitable cost of complexity. Some of it is pure organisational sediment, deposited year after year until nobody remembers whether it was once a bridge, a dam or merely a very expensive puddle.
AI really can do much of this work. It can track workflows, summarise documents, spot anomalies, compare performance, draft reports, enforce procedures, check compliance, notice delays and produce dashboards. It can make an organisation legible to management in real time. It does not get tired. It does not need to be invited to alignment meetings. It does not quietly lose the will to live when asked to produce another quarterly risk summary. From the point of view of the firm, this is extraordinary. The internal nervous system of the company can be automated, accelerated and centralised. A CEO who once depended on successive layers of human interpretation can now see further into the organisation than ever before.
This is precisely where the papal warning begins to bite. Magnifica Humanitas does not deny that technology can relieve human beings of repetitive, dangerous or burdensome work. It does not ask us to return to candlelight, quills and the pious scratching of monastic copyists, pleasant though that might be for those of us with a weakness for vellum. Its concern is rather that efficiency can become a false anthropology. The encyclical warns against reducing human value to usefulness, productivity or measurable output, and it insists that economic and technological systems must remain ordered towards the dignity of the person and the common good.
This is the central clash. Prince sees “measurers” as a category of organisational function. Leo sees workers as persons embedded in families, communities, duties, hopes and fears. Prince asks which roles still create or capture value for the company. Leo asks whether society still grants people a place when the company no longer needs them. Both questions are rational. Only one is sufficient.
The word “measurer” is itself revealing. It sounds modest, almost technical. Yet it gathers together a huge portion of modern middle-class employment: middle management, finance, legal support, compliance, operations, marketing analytics, HR coordination, audit, project management and all the grey connective tissue that holds institutions together. These are not merely jobs. They are mortgages, pensions, school fees, Friday lunches, professional identities, LinkedIn profiles, quiet ambitions, minor vanities and the accumulated promise that education would lead to security. To describe this whole world as “measuring” may be analytically useful, but it is socially explosive.
We have seen this before, though not in exactly this form. The industrial revolution displaced craft workers, rural labourers and whole ways of life. The post-war managerial revolution then created new classes of clerks, administrators, experts and coordinators. Deindustrialisation destroyed many manual occupations, while expanding the white-collar world of offices, credentials and paperwork. For decades, the implied social bargain was that physical labour might decline, but educated administrative labour would grow. The son of the factory worker would not necessarily make things; he might manage processes, analyse data, ensure compliance, design communications strategies or write policy documents. It was not heroic, but it was a life. And now the machines have entered the office.
This is why Douglas Adams’ Golgafrincham joke suddenly feels less like absurd comedy and more like a memo from the future. In The Hitchhiker’s Guide to the Galaxy, a civilisation sends away its supposedly useless middle class on a spaceship: hairdressers, telephone sanitisers, management consultants and others deemed expendable. The remaining population congratulate themselves on having rid society of its dead weight. Then comes the punchline, as cruel and precise as all proper satire should be: the people left behind are later wiped out by a disease spread through dirty telephones, because they had sent away the telephone sanitisers. The joke works because the distinction between useful and useless people is both ridiculous and seductive. Every civilisation is tempted to imagine it can separate the essential from the inessential with surgical clarity. Usually it discovers only later, and usually at some cost, that some of the apparently inessential people were maintaining the conditions that made the obviously essential people effective.
That, perhaps, is the hidden danger in the Cloudflare model. Some measurement is dead bureaucracy. Some is institutional conscience. Some middle management is pointless interference. Some is mentoring, translation, protection, tact and memory. Some compliance work is box-ticking. Some is the thin barrier between ambition and abuse. Some operations work is routine coordination. Some is the art of knowing which rule to bend, which person to call and which problem is not yet visible in the data. The machine may measure what can be measured continuously, but it may also tempt leaders to believe that what cannot be measured continuously does not matter.
This is not an argument for preserving every job exactly as it is. That would be nostalgia disguised as ethics. Many organisations are indeed bloated. Many reporting systems exist chiefly because earlier reporting systems created the need for people to report on the reporting. There is no moral duty to maintain spreadsheet liturgy for its own sake. A civilisation should not force people to spend their lives producing documents nobody reads in order to preserve the fiction of employment. That would be a strangely cruel version of humanism: saving the worker by saving the pointless task.
The most fundamental question, then, is not whether the measurers should all remain in their current posts. Many should not. Some roles will disappear because they were never truly necessary; others because AI can perform them faster, more cheaply and more consistently. But who is responsible for finding new roles for the people whose old roles vanish? If the answer is “each individual, alone”, then we have simply rebuilt Ark B without the inconvenience of constructing an actual spaceship. We will not launch the measurers into space; we will leave them on the labour market, armed with transferable skills, motivational webinars and the faint consolation that their former employers are now more agile.
A humane society cannot treat this as a private sorting exercise. The question of what displaced workers should do next is too large to be left to corporate restructuring teams, recruitment algorithms and the desperate optimism of retraining brochures. If AI removes a large part of the organisational middle, then new forms of contribution must be consciously created: in care, education, public administration, local democracy, culture, environmental repair, civic infrastructure and forms of human mediation that machines can support but should not replace. The point is not to preserve the old measurement class unchanged. It is to prevent a class of people from being told, implicitly or explicitly, that they were useful only until the dashboard improved.
Nor can the answer simply be retraining. Retraining is useful, but it has become the comfort blanket of every politician facing structural change. It allows governments to imply that unemployment is mainly a skills mismatch, and that those displaced by technology can be smoothly converted into tomorrow’s in-demand workers, like files exported from one format to another. Some people will become builders. Some will become sellers. Some will create new kinds of work we cannot yet name. But not everyone can become an elite engineer, a trusted relationship-builder or an entrepreneur of the self, selling personal authenticity through a subscription platform until the sun burns out.
Nor can the answer simply be universal basic income, though some form of income floor may become necessary. Money matters, but money alone does not solve the crisis of place. A society that pays people to be quiet while capital and machines perform the recognised work of civilisation may avoid destitution while deepening humiliation. Universal basic services, shorter working weeks, stronger public institutions, worker ownership, cooperative structures, civic work, cultural work, care work and new forms of democratic participation may all have to become part of the answer. The point is not to preserve every old job, but to prevent economic redundancy from becoming social exile.
This is where Prince and the pope might, unwillingly, need each other. Prince offers a clear description of what AI does inside the firm: it compresses measurement, reduces coordination costs and lets management redirect labour towards production and sales. Leo offers the necessary moral audit: no society can be judged solely by the efficiency of its most successful firms. The company asks whether a role is still necessary. The Church asks whether a person is still recognised. The state, if it remembers its purpose between procurement scandals and digital strategy launches, must ask how these two truths can be reconciled.
A company can rationally remove 20% of its workforce and become stronger. An economy in which many companies do the same at once may become politically unstable, socially cruel and morally incoherent. What is sensible at the level of the individual firm can be disastrous at the level of civilisation. This is one of the oldest problems in political economy, and it has lost none of its charm, by which I mean its capacity to ruin lives while looking elegant on a slide.
That is why the future of the measurers cannot be left entirely to those who no longer need them. Firms have incentives to identify redundancy. They do not have adequate incentives to create new social purposes for the redundant. Markets are good at registering demand backed by money. They are much weaker at recognising latent human capacity, civic need or forms of value that cannot immediately be monetised. A cleaner river, a less lonely old person, a better local archive, a safer pedestrian crossing, a child who learns to read, a public service that answers the phone and actually solves the problem: these are not necessarily the tasks that private capital will discover when it automates compliance reporting. But they are real forms of civilisation.
The Last Revolution will therefore require more imagination than the first industrial revolution demanded. It is not enough to say that new jobs will appear because new jobs have appeared before. That may be true, but it is not a policy. It is a historical shrug. The new roles will have to be invented, funded, legitimised and distributed. Some may emerge in the private sector. Many will not. If AI makes society vastly more productive, then a portion of that productivity must be claimed for common purposes. Otherwise the result will not be liberation from drudgery, but a narrower economy in which a smaller number of humans are richly integrated into production while many others are told to become flexible, resilient and invisible.
The Golgafrincham lesson is not that all apparently useless people are secretly useful in obvious ways. It is subtler and nastier than that. Civilisations are very bad at knowing what kind of human activity they can safely discard. They mistake visibility for value, measurement for truth and efficiency for wisdom. They classify people by function, then forget that functions were attached to lives and systems. If we do not want to build Ark B, the task cannot be merely to identify who is no longer needed. It must be to decide, collectively and deliberately, where those people can next be needed.
That decision cannot be outsourced to AI either. AI may help us see needs, match skills, design institutions and allocate resources more intelligently. It may expose waste, reveal hidden shortages and suggest new forms of public service. But the question of what human beings are for is not a technical question. It is political, moral and civilisational. A dashboard can show that a function has become redundant. It cannot tell us whether a society has become indecent.
The uncomfortable possibility is that both Prince and Leo are right. AI may genuinely allow firms to become more productive by automating vast regions of internal measurement. Many present-day roles may indeed be transitional artefacts of an era when humans were needed to make complex organisations legible to themselves. But Leo is right that the human person must not be reduced to a transitional artefact. The end of a role must not mean the cancellation of a life. The disappearance of a task must not become the disappearance of a citizen.
Non omnia possumus omnes. We cannot all do everything. We cannot all be builders, sellers, founders or technical prodigies. A humane civilisation does not begin by denying that fact. It begins by asking how many different ways there can be to belong. The danger is not that AI will make some forms of work unnecessary. The danger is that we will respond by deciding that the people who did that work are unnecessary too. That is the road to Golgafrincham. And we already know what happened after they launched Ark B.