Today’s radar starts from a shared tension. Institutions try to order complex worlds through standards, rubrics, inventories, ethical principles and automated systems. Yet social, ecological and educational life does not always fit neatly into those moulds.

The anthropological question is not only whether a tool works. It is what it makes legible, what it leaves out and who decides the form that life, justice or truth must take in order to be recognised.

Anna Tsing and fungi against obedient design

In Rotterdam, the exhibition FUNGI. Anarchist Designers, at the Nieuwe Instituut, runs from 20 November 2025 to 8 August 2026. It is curated by anthropologist Anna Tsing and architect and artist Feifei Zhou. The exhibition presents fungi as independent organisms that thrive in multi-species environments and in the ruins of capitalism, rather than as obedient materials for architecture or sustainable design.

The starting point is critical. Tsing and Zhou reject the idea of incorporating fungi into design and production processes as if they were a new friendly raw material. The exhibition defines itself as “anti-design” and presents fungi as anarchic, unconscious and uncontrollable co-designers of worlds that only exist through alliances between human beings and other living beings.

That gesture opens the radar well. Against institutions that need to make life legible in order to manage it, fungi remind us that not everything can be scaled, optimised or turned into infrastructure. The exhibition moves across scales, from the kitchen to the hospital, from termite mounds to coffee plantations, from archives to forests. In an installation on coffee rust in Latin America, the exhibition shows how a fungus can undo an agricultural system built around monoculture and maximum yield.

AIthropology Lab key point — Tsing allows us to begin from what resists standardisation. Not all knowledge is scalable. Not every living practice should become data. Not every relationship between species can be reduced to the language of sustainable innovation. The question is what happens when the living can enter an institution only after becoming representable, safe and administrable.

SARI 2.0 and the automation of the good question

In Chile, the University of Chile published a note on 10 June 2026 about SARI 2.0, an artificial intelligence tool developed at CIAE to review multiple-choice questions in school, university and health assessments. The platform already collaborates with DEMRE and the Education Quality Agency, and plans to incorporate teams linked to CPEIP, EUNACOM, CONACEM and ACHIEEN.

SARI 2.0 detects construction flaws, labels problems and provides recommendations for improvement. The new stage followed the award of a Fondef IT project by ANID and aims to support the preparation of questions across different educational levels and high-stakes standardised tests.

The point is not only that AI is entering education. The point is that it is entering an infrastructure that already orders pathways, access, certifications and opportunities. In Chile, SARI 2.0 does not arrive in a vacuum. It arrives in an admissions system where the PAES organises registration, test taking, results, applications and selection.

A multiple-choice question may seem like a minor object, but it condenses a whole theory of what it means to know something. It defines what can be asked, what counts as a correct answer, which forms of reasoning are rewarded and which experiences remain outside the format.

AI does not create educational inequality by itself. It can, however, optimise an infrastructure that was already classifying people. Technically improving a question does not necessarily make an assessment fairer. A question can be well written and still measure from narrow social, linguistic or territorial assumptions.

This is where the delicate loop appears. If a tool learns to recognise the good question from historical assessment criteria, it can also consolidate the biases of those assessments. What was once a psychometric tradition then becomes an automatic recommendation.

AIthropology Lab key point — SARI 2.0 does not only review items. It participates in stabilising criteria around clarity, validity, error, difficulty and relevance. In high-stakes tests, those criteria can affect university access, professional certification and the recognition of competences. The question is whether improving the technical quality of a question makes assessment fairer, or merely more efficient in its unfairness.

Oaxaca and the city that does not fit the inventory

In Oaxaca, the signal now has a direct institutional source. UNESCO published a note on 26 June 2026 about communities in Oaxaca de Juárez building the city’s first intangible cultural heritage inventory together with UNESCO, INAH, UCUVI and SEDATU. The stated aim is for living heritage to be integrated into urban planning and to guide responsible and sustainable tourism in the hands of its communities.

The source mentions expressions such as the traditional lacework of the Xochimilco neighbourhood, artisanal snow or ice cream, the mayordomía of Xochimilco and the comparsas of Jalatlaco. It also underlines the participation of bearer communities in registration, documentation and knowledge-exchange processes. That frame is promising, but it should not simply be repeated as good practice. The question cannot be limited to celebrating participation. We need to ask how it is produced, who represents it and which voices remain outside.

Yet the fracture appears there too. Sustainable tourism is still tourism. It may redistribute benefits, but it can also turn community life into an economic resource, an urban brand or a consumable experience. Living heritage risks being protected only insofar as it can be narrated, inventoried and visited.

Oaxaca allows us to think about two ways of ordering the city. On the one hand, technocratic planning measures flows, densities, mobility, risk, profitability or tourist appeal. On the other, a form of planning tries to recognise situated knowledge, neighbourhood memories and community practices. The question is not which of the two is pure. The question is what happens when a community must translate its common life into the language of an urban strategy, an international organisation or a tourist economy.

Institutional legibility is gained. Opacity, internal conflict, autonomy and the right not to become a postcard may be lost.

AIthropology Lab key point — Urban planning is not only a technique of ordering. It is a dispute over who can define what counts as city, what counts as heritage and which ways of inhabiting deserve continuity. The question is whether a community can be a protagonist if the recognisable form of participation is defined by an external institution.

UNESCO and AI ethics as a promise of translation

UNESCO’s Recommendation on the Ethics of Artificial Intelligence does not enter here as a piece of current regional news. It enters as a background frame. It was adopted in November 2021 by UNESCO’s 193 Member States and states that AI must respect human rights and human dignity, with principles such as transparency, fairness, environmental sustainability and human oversight.

The framework also refers to multi-stakeholder governance, privacy, auditability, traceability, human oversight, sustainability, public literacy and non-discrimination.

Its limit appears precisely in translation. Ethical principles sound strong in a document, but they only become real when they reach a database, a classroom, a test, an urban licence, a platform or a public decision system.

SARI shows the micro level where an algorithmic policy becomes a technical criterion. Oaxaca shows the territorial level where community participation may be translated into a grammar of planning. Tsing reminds us that there are always forms of life, human and non-human, that resist being fully domesticated.

AIthropology Lab key point — AI ethics is not tested in the declaration. It is tested in the infrastructures that classify, assess, recommend, authorise or exclude. The question is what the value of human-centred AI is if people only appear after the system has already learnt how to measure them.

Cross-reading

We begin with fungi, with Tsing. We move to code, with SARI. We enter the territory, with Oaxaca. And we leave AI ethics as a framework that promises to guide these translations without resolving them by itself.

The pieces do not speak about the same thing, but they share an operation. They show how different worlds are translated in order to become governable.

Tsing’s exhibition reminds us that the living exceeds design. SARI turns assessment into criteria that can be reviewed by a machine. Oaxaca allows us to ask what happens when living heritage enters the language of urban planning. UNESCO offers a global ethical vocabulary, but its effectiveness depends on what happens in concrete infrastructures.

Every institutional translation has costs. To become governable, a practice must become legible. To become legible, it must lose part of its ambiguity. To enter a database, a rubric, an inventory, a display case or an urban plan, it must accept a form.

The anthropological question is not whether the tool works. It is what world it makes visible. Which knowledges it rewards. Which voices must translate themselves in order to be heard. Which forms of care become data, standard, heritage or product.

At bottom, the problem is not technical. It is political and cultural.

Who defines a good question?

Who writes the script of a living city?

And which forms of life resist becoming norm, data or design?

Sources