Batlle y Ordonez against the algorithm
If Batlle y Ordonez appeared today in Montevideo, he would not first ask whether artificial intelligence is intelligent.
He would ask who pays for it.
Who owns it.
Who works under it.
Who is measured by it.
Who can appeal it.
Who captures the productivity it promises.
And who is left outside when the system says no.
The Batllist question would not be whether the machine thinks. It would be who organizes life when the machine begins to decide.
This is not a nostalgic exercise.
It is a political method.
To imagine Batlle before the algorithm is not to bring a bronze statue into the present. It is to ask what a tradition of public intervention, social rights, secularization, education and State-building would do if the new infrastructure of power were computational.
He would not return to open a museum
Batlle would not return to inaugurate a museum of Batllism.
He would not ask us to repeat old slogans while the economy reorganizes itself elsewhere.
He would not confuse loyalty with preservation.
The interesting Batlle is not the one framed in a schoolbook, but the one who understood that politics had to enter the decisive machinery of its time: work, energy, education, public services, citizenship, the relation between capital and labor, the organization of social risk.
If that is the method, then today’s machinery includes data centers, platforms, algorithms, cloud contracts, automated decision systems, AI models, digital identity, public procurement, labor metrics and the privatized infrastructures through which life is increasingly managed.
The question is not whether Batlle would have liked technology.
The question is whether he would have left its power untouched.
Probably not. Batlle did not think the social order was a natural fact to be admired from the balcony. He saw institutions as tools for altering the distribution of power. That is the part worth recovering. Not the exact answers of another century, but the refusal to accept that the strongest private infrastructure should define the terms of citizenship by default.
Batllism was a policy of social infrastructure
Batllism was not only a set of reforms.
It was an attempt to build social infrastructure: eight-hour workday, public services, secular State, education, labor rights, institutional imagination, public companies, a stronger idea of citizenship.
That infrastructure did not eliminate conflict. It organized conflict in a way that made life less dependent on private arbitrariness.
The algorithmic world needs the same kind of question.
What is the twenty-first-century equivalent of social infrastructure?
Not only fiber optics.
Not only connectivity.
Not only digital government.
Also public capacity to understand systems, audit them, negotiate contracts, protect workers, explain decisions, preserve data sovereignty and prevent essential services from becoming black boxes owned elsewhere.
That boring capacity is politically decisive. Without it, sovereignty becomes theatrical. A minister can announce an AI strategy, a public office can launch a digital service, a school can sign a platform agreement, and yet the real knowledge remains outside the State. The country becomes a client of systems it cannot fully inspect.
The social State of the twentieth century had to build institutions around work, health, education and energy.
The social State of the twenty-first has to build institutions around data, computation, explanation and technical dependence.
AI as the new social question
In the early twentieth century, the social question was not solved by telling workers to adapt individually to industrial change.
It required law, institutions, conflict and redistribution.
AI is not the same as industrialization, but it raises a comparable political problem: a new productive force enters existing inequalities and promises efficiency. The question is who receives the gain and who absorbs the risk.
If AI increases productivity, do workers receive shorter hours, better wages, training and more autonomy?
Or do companies receive margin, concentration, surveillance and a new language for old subordination?
If the State buys AI, does it gain capacity?
Or does it rent intelligence from vendors it cannot audit?
If education uses AI, does it deepen judgment?
Or does it produce students trained to outsource difficulty?
That is the social question in algorithmic form.
The answer cannot be left to individual adaptation. A worker cannot negotiate alone with a platform that defines their score. A teacher cannot solve alone the transformation of assessment. A public agency cannot audit alone a global vendor without shared technical standards. The social question appears precisely when individual effort is no longer enough.
That is why the language of “upskilling” is insufficient. Training matters, but training alone can become a way of privatizing the cost of transition. The worker pays with time, anxiety and uncertainty; the company captures the productivity; the State celebrates modernization; the social question remains untouched.
The Batllist instinct would be to ask for the institution behind the training.
Who pays?
Who certifies?
Who benefits?
Who is protected during the transition?
Eight hours against the algorithm
The eight-hour day was not only a labor measure.
It was a civilizational statement: life cannot be entirely swallowed by work.
Today, the algorithm does not always extend the working day by making people stay in the factory. It can intensify the day, measure every gesture, assign tasks, rank performance, predict behavior, distribute shifts, punish slowness and make the worker permanently available.
Twenty-first-century social rights begin where the clock meets the dashboard.
A Batllist response would not ask only whether employment survives.
It would ask what kind of employment survives.
It would treat algorithmic management as a labor issue, not as a neutral software feature. Workers should know when they are being evaluated by automated systems. They should have the right to explanation, appeal, collective bargaining over metrics, paid training when tasks change, and protection against the conversion of productivity into pure pressure.
Eight hours against the algorithm does not mean banning tools.
It means refusing to let tools erase the boundary between life and extraction.
The old clock was visible. The new clock can be hidden inside software. It can appear as response time, productivity score, customer satisfaction metric, automatic scheduling, ranking, heat map, availability expectation. That invisibility makes the labor question harder, not less urgent.
A right that cannot see the dashboard arrives late.
Data as public matter
Data is not oil.
That metaphor is too poor.
Data is closer to infrastructure, memory and power.
It can organize access to services, credit, work, education, health, transport, security and public attention. Whoever controls data flows can decide what becomes visible, what becomes measurable, what becomes governable and what disappears from the table.
The social State of the twenty-first century is not measured only by buildings. It is also measured by the invisible infrastructure that decides who enters, who waits and who is left out.
A Batllist State would not treat public data as a raw material to be handed over casually. It would build technical capacity, legal safeguards, audit institutions, open standards where possible and strong public procurement rules.
It would ask: what data leaves the country? Under what contract? With what reversibility? With what public knowledge gained? What happens when the vendor leaves? Who can inspect the system?
Digital sovereignty is not shouting “sovereignty” over imported platforms.
It is the boring capacity to understand, negotiate, audit and replace them.
The replacement part matters. A State that cannot leave a provider is not a client with strategy; it is a captive. Every public contract for critical systems should be judged by what capacity remains in the country when the contract ends. If nothing remains, the modernization was only rental.
This is not anti-foreign paranoia. A small country will always use foreign technology. The issue is whether it does so as a blind consumer or as a public actor with memory, bargaining power and exit routes. Dependency is not created by importing a tool. Dependency is created by importing a tool in such a way that no one local understands the system, the data, the risks or the alternatives.
Education, not training
Batllism understood education as nation-building, not as narrow job training.
That matters now.
The AI answer cannot be reduced to prompt courses. A country does not become intelligent because thousands of people learn to ask a chatbot for a summary. It needs mathematics, programming, data literacy, philosophy, history, language, art, ethics, public reasoning and technical practice.
Education must produce people who can use tools and question the world those tools build.
That means teaching verification, source criticism, statistical intuition, writing, argument, attention, civic responsibility and the limits of automation. The student should not leave school afraid of AI, but also not enchanted by it.
The goal is not purity.
The goal is judgment.
That judgment has to be democratic, not only expert. Experts are necessary, but public life cannot be reduced to technical priesthood. Citizens need enough understanding to ask better questions; journalists need enough literacy to avoid both panic and propaganda; unions need technical knowledge to bargain; teachers need institutional support to redesign learning rather than merely police cheating.
Education, in that sense, is the first audit institution. A society that cannot understand the systems it uses will end up believing them. It will confuse interface with truth, speed with competence and automation with neutrality. The classroom is where that confusion can be interrupted before it becomes common sense.
Secularizing the platform
Batlle fought for secularization in a country where religious power had to be separated from public authority.
Today, another kind of authority presents itself as natural: platform authority.
The interface says.
The algorithm recommends.
The model ranks.
The dashboard shows.
The system decides.
Those sentences can become a new theology if politics does not interrupt them. Secularizing the platform means refusing to treat technical output as revealed truth. It means asking how it was built, with what data, for whose interest, with what error, with what right to contest.
It also means resisting private vocabularies that enter public life as if they were neutral: engagement, optimization, friction, conversion, productivity, relevance. Behind each word there is a model of the human being.
A minimum program
A minimum Batllist program for AI would not begin with a slogan.
It would begin with institutions.
Public registry of high-impact automated systems used by the State.
Audit rights for labor, education, health, credit, security and public services.
Collective bargaining over algorithmic management.
Public procurement clauses for explainability, data protection, reversibility and knowledge transfer.
Investment in public technical capacity, not only licenses.
Education that joins technical literacy with humanities.
Protection against automated decisions that cannot be appealed.
Tax and social security debate around productivity captured by automated capital.
Support for local research, open standards and public-interest technology.
None of this is anti-innovation.
It is politics refusing to arrive after the infrastructure is already private fact.
A program like this would be accused of slowing innovation. That accusation is predictable. Every social right was, at some point, described as an obstacle by those who benefited from the absence of limits. The question is not whether rules slow something. The question is what kind of acceleration we are accepting without them.
There are accelerations that liberate.
There are accelerations that exhaust.
There are accelerations that make a public service work better.
There are accelerations that make responsibility disappear faster.
Politics begins by distinguishing them.
What he would not do
Batlle would not kneel before Silicon Valley because the vocabulary sounds modern.
He would not confuse deregulation with freedom.
He would not treat workers as friction in a productivity graph.
He would not let public administration buy black boxes and call that modernization.
He would not reduce education to employability.
He would not say that because the machine is new, the old questions of power, labor, ownership and citizenship have expired.
He would also not treat the State as automatically virtuous. A public algorithm can be abusive. A public database can be careless. A public office can buy badly, explain badly and harm people with a stamp of legality. The answer to private power is not blind State power. It is democratic public capacity: transparent, contestable, technically competent and politically accountable.
The counterfactual matters only if it gives us a standard for the present.
The question is not what Batlle would tweet.
The question is whether Uruguay still has the institutional imagination to build public answers when private infrastructure reorganizes common life.
Against the algorithm does not mean against technology.
It means against surrender.
That distinction matters because the easy caricature is always available. Anyone who asks for rights, audits or public capacity can be accused of fearing the future. But the opposite is true. Fear of the future often hides inside passive enthusiasm: the belief that whatever arrives with enough capital and technical vocabulary must be accepted as destiny.
A democratic country cannot afford that innocence.
It has to learn, buy, build, regulate, refuse, negotiate and invent at the same time.
That is the hard part.
The algorithm is not the enemy.
The enemy is a political imagination so poor that it can only choose between worship and panic.
Batlle, if he serves for anything today, serves to reject that poverty.
He also serves to remind us that institutions are not born after consensus. They are built in conflict, against interests that call themselves common sense. AI will have its own common sense: faster, cheaper, inevitable, personalized, frictionless. A democratic politics has to ask the impolite question inside each promise: for whom?
And then a second one: under what institution? Because without institution, even the best technological promise becomes an arrangement of private power and public dependence.
That is the line this counterfactual tries to draw: not a cult of the State, not a cult of the market, but the old republican insistence that the powers organizing common life must be visible, disputable and answerable.
Nothing less deserves to be called modernization.
Everything else is surrender with better lighting.
The name of progress is not enough.
Martín Álvarez
@unfalsoguru