A working vocabulary

A · B · C · D · E · F · G · H · I · L · M · N · O · P · R · S · T · V

Actor-network theoryLatour, Callon, Law, 1980s

Treats humans and non-humans — machines, documents, microbes — as actors in networks, refusing a clean split between the social and the technical. Useful for tracing how an AI system's behaviour is produced by a chain of people, data, code and institutions rather than by the model alone.

Agency and structureGiddens, Bourdieu and many others

The central tension in social theory: how far people make their own choices (agency) and how far those choices are shaped by social arrangements (structure). Giddens's structuration theory argues that structures are both the medium and the outcome of action.

Algorithmic imaginaryTaina Bucher, 2017

The ways people imagine, perceive and feel about algorithms. These beliefs shape behaviour — posting at certain times, using certain words — whether or not they are accurate, and so feed back into the system.

AlienationKarl Marx, 1844

The estrangement of workers from the product of their labour, from the activity of work, from each other, and from their own human potential. Revived in debates about whether AI-assisted work leaves people feeling the output is not really theirs.

AnomieÉmile Durkheim, 1893 and 1897

A condition of normlessness in which old rules have lost their hold and new ones have not formed, leaving people without guidance about what to want or how to behave. The concept most often reached for in moments of rapid technological change.

AsabiyyaIbn Khaldun, 1377

Group feeling or social cohesion. In Ibn Khaldun's theory, strong asabiyya lets a group seize power; luxury and dependence on hired agents erode it, and the dynasty falls within a few generations.

Audit societyMichael Power, 1997

A society in which checking and verification spread into every institution. Power warned that audits can become rituals that generate reassurance rather than real accountability — a live risk for algorithmic audits.

Automation biasHuman factors research, 1990s

The tendency to over-rely on automated recommendations, accepting them without sufficient scrutiny and failing to notice when they are wrong. A key reason why a nominal “human in the loop” may not provide real oversight.

Banality of evilHannah Arendt, 1963

Arendt's phrase for Adolf Eichmann, who appeared not as a monster but as a careerist functionary. The idea that great harm can be done by ordinary people following procedures has shaped thinking about responsibility in bureaucratic and automated systems.

Biographical availabilityDoug McAdam, 1986

The absence of personal constraints, such as a full-time job, a spouse or children, that would make the costs of activism higher. McAdam found it, along with social ties, predicted who went to Mississippi for Freedom Summer in 1964.

Blasé attitudeGeorg Simmel, 1903

The reserved, indifferent stance city dwellers develop to cope with an overload of stimuli. A good description of how people manage endless feeds and notifications.

BureaucracyMax Weber

Administration through a hierarchy of offices, written rules, impersonal procedures and appointment by expertise. Weber saw it as technically superior and deeply dangerous. Algorithmic systems are often analysed as an extension of bureaucratic logic.

Captive mindSyed Hussein Alatas, early 1970s

A mind dominated by Western thought, uncritically imitating its concepts and problems. Alatas applied it to scholars in post-colonial societies; it now applies to AI systems trained mostly on Western text.

Charismatic authorityMax Weber

Authority based on devotion to the exceptional qualities of a leader rather than on tradition or rules. Weber thought it unstable and destined to be “routinised” into tradition or bureaucracy.

Classification situationMarion Fourcade and Kieran Healy, 2013

A position in society defined by one's place in algorithmic rankings and scores, analogous to Weber's “class situation” defined by market position. Developed in The Ordinal Society (2024).

Co-productionSheila Jasanoff, 2004

The idea that scientific and technological knowledge and social order are produced together, each shaping the other. A corrective to both technological determinism and the view that technology is merely a neutral tool.

Collective effervescenceÉmile Durkheim, 1912

The charged energy a group feels when it gathers and acts together in ritual, which Durkheim saw as the source of the sacred. Visible in stadiums, protests and viral moments.

Collective efficacyRobert Sampson and colleagues, 1997

Mutual trust among neighbours combined with a willingness to intervene for the common good. Sampson's Chicago research found it a strong predictor of lower violence, more so than visible disorder.

Coloniality of powerAníbal Quijano, 2000

The argument that racial classification and the global division of labour were created together in the conquest of the Americas, and that these hierarchies outlived formal colonialism.

Cultural capitalPierre Bourdieu

Knowledge, skills, tastes and credentials — much of it acquired at home — that confer advantage, especially in school. One of Bourdieu's forms of capital, alongside economic, social and symbolic capital.

Cultural omnivoreRichard Peterson, 1992

The finding that high-status people in the late twentieth century consumed a broad range of culture, high and popular, rather than only elite forms. Some argue omnivorousness is itself a new marker of distinction.

Culture industryMax Horkheimer and Theodor Adorno, 1944

Mass-produced culture that standardises taste and encourages passivity. The ancestor of critiques of engagement-maximising platforms.

Data colonialismNick Couldry and Ulises Mejias, 2019

The appropriation of human life through data, framed as a new phase of colonial extraction rather than merely a metaphor. Contested, but influential in Global South debates about AI.

Differential mode of associationFei Xiaotong, 1947

Fei's description of Chinese rural social structure, in which each person is the centre of concentric rings of relationships, like ripples from a stone, with obligations weakening with distance.

Digital divide1990s onward

Unequal access to digital technology. Researchers now distinguish a first-level divide (access), a second level (skills and use) and a third (who benefits). Generative AI adds a fourth: who uses AI as a tool and who is subject to it.

Digital poorhouseVirginia Eubanks, 2018

Eubanks's term for systems of automated eligibility, risk scoring and data collection that subject poor people to intense surveillance and suspicion that others do not face.

DisenchantmentMax Weber, 1917

The decline of magic and mystery in a world understood through science and calculation. Ironically, people often re-enchant AI, attributing to it understanding or even personality.

Double consciousnessW. E. B. Du Bois, 1903

The sense of always looking at oneself through the eyes of a contemptuous world, which Du Bois described as the experience of Black Americans. Now also applied to living under algorithmic classification.

Double movementKarl Polanyi, 1944

The pattern by which the expansion of markets provokes a counter-movement of social protection: labour laws, unions, regulation. AI regulation is increasingly read as a new counter-movement.

DramaturgyErving Goffman, 1956

The analysis of social life as performance, with front stages where people manage impressions and back stages where they relax. Applied to online self-presentation and to performances aimed at algorithms.

Economic connectednessRaj Chetty and colleagues, 2022

The share of high-income friends among people with low incomes. Measured from Facebook data on 72 million people, it was one of the strongest predictors of upward mobility ever found.

ELIZA effectAfter Joseph Weizenbaum's 1966 program

The tendency to attribute understanding, empathy or intelligence to computer systems that imitate conversation. Weizenbaum was alarmed by how readily people confided in a very simple program.

Emotional labourArlie Russell Hochschild, 1983

The management of one's own feelings to produce a required emotional state in others, as part of paid work. Raises questions both about workers replaced by chatbots and about machines that perform care.

Fictitious commoditiesKarl Polanyi, 1944

Labour, land and money: things treated as commodities by market society although they were not produced for sale. Polanyi argued that fully commodifying them destroys the social fabric. Some now add personal data and creative work.

FieldPierre Bourdieu

A structured social arena — science, art, politics, journalism — with its own rules, stakes and forms of capital, in which actors compete for position.

Filter bubbleEli Pariser, 2011

The idea that personalisation algorithms isolate users in information environments that confirm their existing views. Popular, but the empirical evidence for strong filter bubbles is weaker than the idea's fame.

General intellectKarl Marx, 1857–58

Society's accumulated knowledge and skill, which Marx imagined becoming embodied in machines as the main force of production. Frequently cited in discussions of AI models trained on the collective output of humanity.

Ghost workMary Gray and Siddharth Suri, 2019

The hidden human labour — data labelling, moderation, verification — that makes automated systems work, often done by contract workers paid per task.

HabitusPierre Bourdieu

A set of deeply ingrained dispositions, tastes and ways of perceiving acquired through upbringing and social position, which feel natural to the person who has them.

Hegemonic masculinityRaewyn Connell, 1980s

The culturally dominant ideal of manhood in a given setting, which legitimises men's dominance over women and over other, subordinated masculinities.

HegemonyAntonio Gramsci, 1929–35

Rule by consent rather than force, achieved when a dominant group's worldview becomes common sense. Used by left and right alike, and relevant to whoever shapes what AI systems treat as neutral.

HomophilyLazarsfeld and Merton, 1954; McPherson et al., 2001

The tendency for people to form ties with others like themselves. It makes social influence hard to distinguish from selection in network data.

Ideal typeMax Weber

An analytical model that exaggerates key features of a phenomenon so it can be compared with reality. Weber's bureaucracy and types of authority are ideal types, not descriptions.

IntersectionalityKimberlé Crenshaw, 1989

The insight that systems of inequality such as race and gender combine in ways that cannot be understood by examining each separately. Made concrete in AI audits that check error rates across combined groups.

Invisible cageHatim Rahman, 2021

Rahman's term for the situation of platform workers controlled by opaque, frequently changing rating algorithms whose rules they cannot see.

Iron cageMax Weber, 1905

Talcott Parsons's translation of Weber's image of a rationalised economic order that, once established, traps people in calculation and routine. Widely invoked for algorithmic systems.

Iron law of oligarchyRobert Michels, 1911

The claim that every organisation, however democratic in intent, comes to be dominated by a small leadership. “Who says organisation, says oligarchy.”

Labelling theoryHoward Becker, 1963, and others

The view that deviance is not a quality of an act but the result of others applying a label to it, which can then shape the labelled person's identity and opportunities.

Latent deprivationMarie Jahoda, 1982

Jahoda's theory, rooted in the Marienthal study, that employment provides time structure, social contact, collective purpose, status and activity, and that losing these harms well-being independently of income.

LegitimacyMax Weber

The belief that a form of authority is rightful and its commands binding. Weber distinguished traditional, charismatic and legal-rational legitimacy. A central question for decisions made by machines.

Liar's dividendBobby Chesney and Danielle Citron, 2019

The benefit liars gain when the public knows convincing fakes exist: genuine evidence can be dismissed as fabricated.

Liquid modernityZygmunt Bauman, 2000

Bauman's term for a condition in which social forms, relationships and identities no longer hold their shape long enough to become settled.

Looping effectIan Hacking, 1995

The way classifications of people change the people classified, who respond to the classification, which in turn changes the category. Machine scoring creates looping effects at enormous scale.

Matthew effectRobert K. Merton, 1968

Cumulative advantage: those who already have recognition get more of it. Named after the Gospel of Matthew. Recommendation systems that rank by popularity tend to produce it.

McDonaldizationGeorge Ritzer, 1993

The spread of fast-food restaurant principles — efficiency, calculability, predictability, and control through technology — to more and more sectors of society. An update of Weber's rationalisation.

Mechanical and organic solidarityÉmile Durkheim, 1893

Two bases of social cohesion: likeness and shared beliefs in simple societies (mechanical) and interdependence through the division of labour in complex ones (organic).

Model collapseIlia Shumailov and colleagues, 2024

The degradation that occurs when AI models are trained recursively on data produced by earlier models, progressively losing information about rarer parts of the original data.

Moral crumple zoneMadeleine Clare Elish, 2019

The tendency for the human operator in an automated system to absorb blame for failures, even when they had limited control, much as a car's crumple zone absorbs a crash.

Moral panicStanley Cohen, 1972

An episode in which a group or behaviour is defined as a threat to social values, with exaggerated media attention and calls for control. Cohen's study was of Mods and Rockers; many technology scares share its pattern, which does not mean every concern is a panic.

Network societyManuel Castells, 1996

A social structure organised around digital information networks, in which power lies in programming and switching between networks.

New Jim CodeRuha Benjamin, 2019

Discriminatory technologies that encode inequity while appearing neutral or objective, named after the Jim Crow laws of racial segregation in the US.

OperationalisationResearch methods

The translation of an abstract concept, such as health need, intelligence or risk, into something measurable. Many AI harms come from a gap between the concept that matters and the proxy that is measured.

PanopticonJeremy Bentham, 1791; Michel Foucault, 1975

Bentham's prison design in which inmates could always be seen but never know when. For Foucault, a model of disciplinary power in which people internalise surveillance. Overused for digital surveillance, which often works because people forget about it.

PerformativityMichel Callon, Donald MacKenzie

The idea that economic and predictive models do not merely describe the world but change it to fit them. MacKenzie showed this for option pricing; AI predictions can act the same way.

PostmaterialismRonald Inglehart, 1977

A shift in values, among generations raised in security, from economic and physical security towards self-expression, autonomy and quality of life.

Preference falsificationTimur Kuran, 1995

Publicly expressing views one does not privately hold, out of social pressure. It explains why revolutions can surprise everyone, including the people who make them.

Public sphereJürgen Habermas, 1962

A space in which private individuals come together to discuss matters of common concern and form public opinion. A standard against which social media and AI-mediated communication are judged.

RationalisationMax Weber

The long-term process by which calculation, efficiency and formal rules displace tradition, emotion and values as the basis of action.

Reflexive modernisationUlrich Beck, Anthony Giddens

A phase of modernity in which society must confront the side effects of its own institutions and technologies.

Relative deprivationSamuel Stouffer and colleagues, 1949

Dissatisfaction based on comparison with others or with expectations rather than on absolute conditions. Found among US soldiers in The American Soldier.

Risk societyUlrich Beck, 1986

A society organised around the production and distribution of manufactured risks, often invisible and global, rather than wealth alone. The framing of the EU AI Act owes much to this line of thought.

Self-fulfilling prophecyRobert K. Merton, 1948

A false definition of a situation that evokes behaviour which makes it come true. Predictive systems that guide action can produce their own confirmation.

Silicon sampleLisa Argyle and colleagues, 2023

The use of a language model, prompted with demographic profiles, to simulate survey responses. Promising for piloting; unreliable as a substitute for people.

Social capitalBourdieu, Coleman, Putnam

The resources that flow through social networks and norms of trust. For Bourdieu, a source of privilege; for Coleman and Putnam, also a public good.

Social construction of realityPeter Berger and Thomas Luckmann, 1966

The process by which human activity produces institutions that come to seem objective and natural, and are then internalised by new generations.

Social factÉmile Durkheim, 1895

A way of acting, thinking or feeling that is external to individuals and exerts constraint on them, such as law, language or a suicide rate. Durkheim insisted they be explained by other social facts.

Sociological imaginationC. Wright Mills, 1959

The ability to connect personal troubles to public issues, biography to history. The single most useful habit sociology teaches.

Standpoint theoryDorothy Smith, Sandra Harding, Patricia Hill Collins

The argument that knowledge is shaped by social position, and that starting from the lives of marginalised groups can reveal what dominant perspectives miss.

Strength of weak tiesMark Granovetter, 1973

The finding that acquaintances, who connect us to other social circles, often provide more new information, including about jobs, than close friends.

Surveillance capitalismShoshana Zuboff, 2019

An economic logic in which firms extract behavioural data beyond what is needed for a service and use it to predict and shape behaviour for profit.

Symbolic interactionismGeorge Herbert Mead; named by Herbert Blumer, 1937

A perspective that sees society as built from interactions in which people interpret and negotiate meanings, and the self as formed by learning to see oneself as others do.

Tactical freezeZeynep Tufekci, 2017

The inability of rapidly mobilised, digitally networked movements to change tactics or make collective decisions, because they skipped the slow organising that builds such capacities.

Thomas theoremW. I. Thomas and Dorothy Swaine Thomas, 1928

“If men define situations as real, they are real in their consequences.”

Threshold modelMark Granovetter, 1978

A model in which each person joins collective action once enough others have. Small differences in the distribution of thresholds can determine whether a riot or a revolution happens or nothing at all.

Total institutionErving Goffman, 1961

A place where all aspects of life are conducted under one authority, cut off from wider society: prisons, asylums, barracks, monasteries.

Two-step flowPaul Lazarsfeld and colleagues, 1944; Katz and Lazarsfeld, 1955

The finding that media influence often passes through opinion leaders who relay it in conversation. The conceptual ancestor of the influencer.

VerstehenMax Weber

Interpretive understanding: grasping the subjective meaning that actions have for those performing them, as part of explaining them.