When we talk about artificial intelligence, we often start too late. We look at the model, its capabilities, the company that develops it, or the data used to train it. Yet before an algorithm classifies, predicts, or generates anything, there is already a world organized through categories, institutions, borders, archives, inequalities, expectations, and ways of imagining the future.

Anthropology does not only study AI. It also studies the world we are beginning to automate.

From universities in the United States, five trajectories in social and cultural anthropology help make that world visible. Arjun Appadurai, Akhil Gupta, Whitney Battle-Baptiste, Setha Low, and Aihwa Ong do not form a single school and they do not share one research object. Not all of them study artificial intelligence directly.

That is precisely what makes bringing them together useful. Their work helps us tune into five frequencies of the present — affect, labor, memory, space, and citizenship — and see how technological infrastructures rest on social relationships that were already there.

When these frequencies overlap, one concern becomes difficult to ignore.

Which social relationships are we turning into technological infrastructure?

Arjun Appadurai — affect and imagination

In April 2026, Arjun Appadurai gave a lecture at the Amsterdam Institute for Social Science Research on the relationship between artificial intelligence, emotion, and democratic imagination. Affect in Times of AI: Emotion, Technology, and Democratic Imagination placed AI systems within a broader transformation of political life, the senses, affect, and the ways solidarity is built.

This move extends questions that have been present in his work for decades. Appadurai showed that globalization never consisted simply of things moving around the planet. People, technologies, images, capital, and ideas circulate at different speeds, collide with one another, and acquire different meanings as they move through different cultural contexts.

Artificial intelligence can now be understood as part of those flows. A model may be developed in the United States, trained on materials produced across many societies, supported by infrastructure distributed across several territories, and eventually embedded in a school in Nairobi, a government office in Madrid, or a phone in Delhi.

The same technology does not arrive in the same world.

His work helps expand the discussion of AI beyond productivity and information processing. Conversational systems can also shape trust, dependency, the imagination of possible futures, and the ways public narratives are formed.

Radar signal. AI is also becoming an affective infrastructure.

What kind of collective imagination emerges when machines begin to participate in it?

Akhil Gupta — labor and unequal futures

In 2025, Akhil Gupta and Purnima Mankekar published The Future of Futurity: Affective Capitalism and Potentiality in a Global City. The book grows out of long-term ethnographic research among people working in call centers and the business process outsourcing industry in Bengaluru, India, largely serving customers in the Global North.

The visible object is the call center. The deeper subject is the future.

Gupta and Mankekar study how people working in this economy organize aspirations, intimacy, consumption, mobility, and expectation around jobs deeply connected to global capitalism. Technology appears not only as a productive tool. It also organizes time, daily rhythms, and horizons of possibility.

This research extends a concern running through Gupta’s work since Red Tape. Institutions and infrastructures distribute possibilities unevenly. Digitization does not necessarily dissolve those relations. It can reorganize them, hide them behind new interfaces, or reinforce structures that were already in place.

That perspective matters as conversational agents begin to take on parts of the work that for decades was outsourced to service centers. An assessment focused only on efficiency or cost savings leaves out local economies, labor trajectories, and the aspirations of people whose lives depended on those jobs.

Radar signal. Automation also reorganizes who gets to imagine what kind of future.

Which social futures are we automating along with that work?

Whitney Battle-Baptiste — memory and data

Whitney Battle-Baptiste directs the W.E.B. Du Bois Center at UMass Amherst and is working on a second edition of Black Feminist Archaeology. Her research connects Black feminist theory, the African diaspora, race, gender, class, material culture, heritage, and historical archaeology.

Her presence in this Radar brings a basic insight to any discussion of data.

Data has a past.

Battle-Baptiste and Britt Rusert brought renewed attention to the visualizations created by Du Bois and his team to represent the conditions and achievements of Black Americans at the 1900 Paris Exposition in W.E.B. Du Bois’s Data Portraits.

Those graphics were not neutral representations. They were a political intervention through data, a way to contest dominant portrayals of Black life produced within racist structures of knowledge.

Battle-Baptiste continues to work on the relationship between materiality, memory, and systemic racism. In February 2026, she returned to the continuing relevance of Du Bois for interpreting the present and to efforts to preserve and activate that legacy.

For artificial intelligence, this shift is crucial. Before evaluating whether a dataset represents a society well, we need to reconstruct how that society came to be recorded in that form. Archives contain presence, but they also contain absence. They preserve what certain institutions considered worth recording, the categories used to record it, and the power relations that determined who could name whom.

Radar signal. There is no data without a history of how that data was produced.

Who historically had the power to produce the archive from which the machine now learns?

Setha Low — space and surveillance

Setha Low has spent decades showing that space is never only where social relationships happen. Space also produces them.

Gated communities, plazas, parks, beaches, homes, and systems of access distribute encounters, separations, rights, and possibilities for participation. In 2024, Low and Mark Maguire pushed this perspective further with Trapped: Life under Security Capitalism and How to Escape It, published by Stanford University Press.

The concept of security capitalism describes an economic, political, and affective ecosystem in which the pursuit of protection turns neighborhoods, airports, cities, and states into increasingly fortified, privatized, and controlled spaces.

That process is not driven only by security companies or state institutions. It is also sustained by everyday fear and by the desire to feel safe.

Low continues to study private governance in New York cooperatives and condominiums and the extension of corporate logics into domestic and public space. In May 2026, she also took part in a CUNY discussion on hostile architecture and democratic design.

The connection to AI is direct. Facial recognition, automated video analysis, biometric controls, anomaly detection, and predictive systems do not operate only inside computers. They produce space. They shape who moves through without friction, whose body triggers additional scrutiny, what behavior activates an alert, and which forms of presence are treated as normal.

Radar signal. The algorithm is acquiring a geography.

What kind of city emerges when algorithmic suspicion is built into its infrastructure?

Aihwa Ong — citizenship, technology, and fungible life

Aihwa Ong is Professor Emerita of sociocultural anthropology at UC Berkeley. Her institutional profile continues to identify science, technology, medicine, knowledge, citizenship, neoliberalism, and urban governance among her areas of research.

Her concept of flexible citizenship showed that citizenship does not operate only as a stable legal relationship between a person and a state. Capital, expertise, mobility, and opportunity allow some groups to negotiate belonging in ways that are not available to everyone.

Her later research on biotechnology showed something similar in another domain. Life itself can acquire different values within global networks of science, capital, experimentation, and knowledge.

This perspective helps us think about AI without treating it as an outside force arriving to transform an otherwise stable society. Technology enters worlds where mobility, access, knowledge, and decision-making power are already unevenly distributed.

The same system may expand possibilities for some people while becoming a new border for others. It can increase an organization’s power to decide while reducing the agency of those who depend on it. It can make some forms of knowledge fungible while dramatically increasing the value of other forms.

Radar signal. Technology enters worlds where power is already distributed.

What new forms of citizenship, authority, and human value emerge when code increasingly participates in deciding who can do what?

When the five frequencies overlap

The five trajectories begin in different places. Appadurai examines flows and imagination. Gupta focuses on labor, institutions, and futures. Battle-Baptiste connects memory and materiality. Low studies space and security. Ong analyzes citizenship, knowledge, and life.

When those frequencies overlap, three larger signals appear.

The myth of the empty world. Before a machine classifies, categories already exist. Before it predicts behavior, institutions have already decided which behaviors matter. Before it generates an image, there is already an archive. Before it monitors a space, there is already a political definition of security. Before it grants access, citizenship and resources are already unevenly distributed.

Infrastructure produces subjectivity. A call center organizes aspiration. A gated community organizes fear. An archive organizes memory. A border organizes citizenship. A platform organizes imagination. AI can also become infrastructure around which people reorganize practices, expectations, relationships, and desires.

The future is not distributed equally. Some people can use new technologies to expand their mobility while others encounter them as systems of control. Some institutions can turn vast cultural archives into computational capital while some communities barely made it into the archive at all. Some people can buy security while others become its object. Some forms of labor can be automated while organizations capture the value produced by that automation.

Five questions can accompany any anthropological reading of an AI system.

  • Before correcting a dataset, do we know which institutions made that archive possible and what was left out?
  • What forms of trust, dependency, and imagination emerge from everyday interaction with conversational systems?
  • Which aspirations, labor trajectories, and local economies disappear from view when productivity is the only thing measured?
  • Which bodies, behaviors, and spaces become suspicious when security is automated?
  • Whose possibilities are expanded by a technology, and for whom does it become a new border?

Anthropology has something important to contribute to the debate on artificial intelligence because it can follow the relationships that precede technology, trace how those relationships change, and notice what disappears when institutions turn the world into categories that machines can read.

What world are we teaching machines to recognize?

What world are we building around their answers?

Relationships come before the algorithm. They remain after it, too.

References