As diversity increases, people paradoxically perceive social groups as more similar

Out-group homogeneity

The out-group homogeneity effect is the perception of out-group members as more similar to one another than are in-group members, i.e. "they are alike; we are diverse".[1] Perceivers tend to have impressions about the diversity or variability of group members around those central tendencies or typical attributes of those group members. Thus, outgroup stereotypicality judgments are overestimated, supporting the view that out-group stereotypes are overgeneralizations.[2] The term "outgroup homogeneity effect", "outgroup homogeneity bias" or "relative outgroup homogeneity" have been explicitly contrasted with "outgroup homogeneity" in general,[3] the latter referring to perceived outgroup variability unrelated to perceptions of the ingroup.

Analytic racecraft: Race-based averages create illusory group differences in perceptions of racism.

Differences Between Tight and Loose Cultures: A 33-Nation Study

The differences across cultures in the enforcement of conformity may reflect their specific histories. , With data from 33 nations, we illustrate the differences between cultures that are tight (have many strong norms and a low tolerance of deviant behavior) versus loose (have weak social norms and a high tolerance of deviant behavior). Tightness-looseness is part of a complex, loosely integrated multilevel system that comprises distal ecological and historical threats (e.g., high population density, resource scarcity, a history of territorial conflict, and disease and environmental threats), broad versus narrow socialization in societal institutions (e.g., autocracy, media regulations), the strength of everyday recurring situations, and micro-level psychological affordances (e.g., prevention self-guides, high regulatory strength, need for structure). This research advances knowledge that can foster cross-cultural understanding in a world of increasing global interdependence and has implications for modeling cultural change.

The homogenizing effect of large language models on human expression and thought

Cognitive diversity, reflected in variations of language, perspective, and reasoning, is essential to creativity and collective intelligence. This diversity is rich and grounded in culture, history, and individual experience. Yet, as large language models (LLMs) become deeply embedded in people’s lives, they risk standardizing language and reasoning. We synthesize evidence across linguistics, psychology, cognitive science, and computer science to show how LLMs reflect and reinforce dominant styles while marginalizing alternative voices and reasoning strategies. We examine how their design and widespread use contribute to this effect by mirroring patterns in their training data and amplifying convergence as all people increasingly rely on the same models across contexts. Unchecked, this homogenization risks flattening the cognitive landscapes that drive collective intelligence and adaptability.

The homogenizing effect of large language models on human expression and thought

Cognitive diversity, reflected in variations of language, perspective, and reasoning, is essential to creativity and collective intelligence. This diversity is rich and grounded in culture, history, and individual experience. Yet, as large language models (LLMs) become deeply embedded in people’s lives, they risk standardizing language and reasoning. We synthesize evidence across linguistics, psychology, cognitive science, and computer science to show how LLMs reflect and reinforce dominant styles while marginalizing alternative voices and reasoning strategies. We examine how their design and widespread use contribute to this effect by mirroring patterns in their training data and amplifying convergence as all people increasingly rely on the same models across contexts. Unchecked, this homogenization risks flattening the cognitive landscapes that drive collective intelligence and adaptability.

The homogenizing effect of large language models on human expression and thought

AbstractCognitive diversity, reflected in variations of language, perspective, and reasoning, is essential to creativity and collective intelligence. This diversity is rich and grounded in culture, history, and individual experience. Yet, as large language models (LLMs) become deeply embedded in people's lives, they risk standardizing language and reasoning. We synthesize evidence across linguistics, psychology, cognitive science, and computer science to show how LLMs reflect and reinforce dominant styles while marginalizing alternative voices and reasoning strategies. We examine how their design and widespread use contribute to this effect by mirroring patterns in their training data and amplifying convergence as all people increasingly rely on the same models across contexts. Unchecked, this homogenization risks flattening the cognitive landscapes that drive collective intelligence and adaptability.

The shrinking landscape of linguistic diversity in the age of large language models

Language is far more than a communication tool; it encodes a wealth of information about a person’s identity, psychological state and social context, providing valuable insights for diverse fields including psychology, marketing and healthcare. Across three studies spanning seven datasets in different domains and over 880,000 texts, we show that the widespread adoption of large language models (LLMs) as writing assistants is linked to declines in linguistic diversity, interfering with the societal and psychological insights language provides. While core content is retained when LLMs polish and rewrite texts, LLMs also homogenize writing styles, reducing writing-complexity variance by a statistically significant 21–50% across datasets and models (P ≤ 0.05), and amplify patterns associated with dominant characteristics while suppressing others, emphasizing conformity over individuality. These trends hold across different LLMs, prompts and contexts, with potential implications for diagnostic processes, personalization efforts, hiring assessments and cultural preservation.

Diversification bias: Explaining the discrepancy in variety seeking between combined and separated choices.

Credit Access in the United States

We measure differences in US households’ access to credit and explore the mechanisms driving such differences using newly constructed population-level linked credit bureau and Census data. We find large differences in credit scores by race, class, and hometown that emerge in one’s 20s and persist throughout the life cycle. These gaps are primarily driven by differences in delinquencies that emerge in young adulthood. By age 30, 73% of Black individuals, 62% of those from low-income families, and 51% of those from Appalachia and the South have a 90+ day delinquency on their credit report, in contrast to 36% for White individuals, 20% for high-income families, and 31% for those from the upper Midwest. These delinquencies are correlated with income and wealth, but observed income profiles and wealth account for at most 10–35% of the gaps in delinquencies across groups. In contrast, movers-based estimates of hometown effects imply that childhood exposure accounts for around 50% of the differences in delinquencies across hometowns. Counties that promote repayment also promote upward income mobility, but adult income mediates only a small fraction of this relationship: growing up in a place where others are likely to repay improves credit outcomes even for those who do not have higher income in adulthood. We provide suggestive evidence on the mechanisms driving these patterns. JEL Codes: G5, H0.

Picking on the Same Person: Does Algorithmic Monoculture lead to Outcome Homogenization?

As the scope of machine learning broadens, we observe a recurring theme of algorithmic monoculture: the same systems, or systems that share components (e.g. datasets, models), are deployed by multiple decision-makers. While sharing offers advantages like amortizing effort, it also has risks. We introduce and formalize one such risk, outcome homogenization: the extent to which particular individuals or groups experience the same outcomes across different deployments. If the same individuals or groups exclusively experience undesirable outcomes, this may institutionalize systemic exclusion and reinscribe social hierarchy. We relate algorithmic monoculture and outcome homogenization by proposing the component sharing hypothesis: if algorithmic systems are increasingly built on the same data or models, then they will increasingly homogenize outcomes. We test this hypothesis on algorithmic fairness benchmarks, demonstrating that increased data-sharing reliably exacerbates homogenization and individual-level effects generally exceed group-level effects. Further, given the current regime in AI of foundation models, i.e. pretrained models that can be adapted to myriad downstream tasks, we test whether model-sharing homogenizes outcomes across tasks. We observe mixed results: we find that for both vision and language settings, the specific methods for adapting a foundation model significantly influence the degree of outcome homogenization. We also identify societal challenges that inhibit the measurement, diagnosis, and rectification of outcome homogenization in deployed machine learning systems.

AI generates covertly racist decisions about people based on their dialect

Hundreds of millions of people now interact with language models, with uses ranging from help with writing1,2 to informing hiring decisions3. However, these language models are known to perpetuate systematic racial prejudices, making their judgements biased in problematic ways about groups such as African Americans4–7. Although previous research has focused on overt racism in language models, social scientists have argued that racism with a more subtle character has developed over time, particularly in the United States after the civil rights movement8,9. It is unknown whether this covert racism manifests in language models. Here, we demonstrate that language models embody covert racism in the form of dialect prejudice, exhibiting raciolinguistic stereotypes about speakers of African American English (AAE) that are more negative than any human stereotypes about African Americans ever experimentally recorded. By contrast, the language models’ overt stereotypes about African Americans are more positive. Dialect prejudice has the potential for harmful consequences: language models are more likely to suggest that speakers of AAE be assigned less-prestigious jobs, be convicted of crimes and be sentenced to death. Finally, we show that current practices of alleviating racial bias in language models, such as human preference alignment, exacerbate the discrepancy between covert and overt stereotypes, by superficially obscuring the racism that language models maintain on a deeper level. Our findings have far-reaching implications for the fair and safe use of language technology.

AI generates covertly racist decisions about people based on their dialect

Hundreds of millions of people now interact with language models, with uses ranging from help with writing1,2 to informing hiring decisions3. However, these language models are known to perpetuate systematic racial prejudices, making their judgements biased in problematic ways about groups such as African Americans4–7. Although previous research has focused on overt racism in language models, social scientists have argued that racism with a more subtle character has developed over time, particularly in the United States after the civil rights movement8,9. It is unknown whether this covert racism manifests in language models. Here, we demonstrate that language models embody covert racism in the form of dialect prejudice, exhibiting raciolinguistic stereotypes about speakers of African American English (AAE) that are more negative than any human stereotypes about African Americans ever experimentally recorded. By contrast, the language models’ overt stereotypes about African Americans are more positive. Dialect prejudice has the potential for harmful consequences: language models are more likely to suggest that speakers of AAE be assigned less-prestigious jobs, be convicted of crimes and be sentenced to death. Finally, we show that current practices of alleviating racial bias in language models, such as human preference alignment, exacerbate the discrepancy between covert and overt stereotypes, by superficially obscuring the racism that language models maintain on a deeper level. Our findings have far-reaching implications for the fair and safe use of language technology.

Can names shape facial appearance?

Our given name is a social tag associated with us early in life. This study investigates the possibility of a self-fulfilling prophecy effect wherein individuals’ facial appearance develops over time to resemble the social stereotypes associated with given names. Leveraging the face–name matching effect, which demonstrates an ability to match adults’ names to their faces, we hypothesized that individuals would resemble their social stereotype (name) in adulthood but not in childhood. To test this hypothesis, children and adults were asked to match faces and names of children and adults. Results revealed that both adults and children correctly matched adult faces to their corresponding names, significantly above the chance level. However, when it came to children’s faces and names, participants were unable to make accurate associations. Complementing our lab studies, we employed a machine-learning framework to process facial image data and found that facial representations of adults with the same name were more similar to each other than to those of adults with different names. This pattern of similarity was absent among the facial representations of children, thereby strengthening the case for the self-fulfilling prophecy hypothesis. Furthermore, the face–name matching effect was evident for adults but not for children’s faces that were artificially aged to resemble adults, supporting the conjectured role of social development in this effect. Together, these findings suggest that even our facial appearance can be influenced by a social factor such as our name, confirming the potent impact of social expectations.

Thinking through other minds: A variational approach to cognition and culture

The processes underwriting the acquisition of culture remain unclear. How are shared habits, norms, and expectations learned and maintained with precision and reliability across large-scale sociocultural ensembles? Is there a unifying account of the mechanisms involved in the acquisition of culture? Notions such as “shared expectations,” the “selective patterning of attention and behaviour,” “cultural evolution,” “cultural inheritance,” and “implicit learning” are the main candidates to underpin a unifying account of cognition and the acquisition of culture; however, their interactions require greater specification and clarification. In this article, we integrate these candidates using the variational (free-energy) approach to human cognition and culture in theoretical neuroscience. We describe the construction by humans of social niches that afford epistemic resources called cultural affordances. We argue that human agents learn the shared habits, norms, and expectations of their culture through immersive participation in patterned cultural practices that selectively pattern attention and behaviour. We call this process “thinking through other minds” (TTOM) – in effect, the process of inferring other agents’ expectations about the world and how to behave in social context. We argue that for humans, information from and about other people's expectations constitutes the primary domain of statistical regularities that humans leverage to predict and organize behaviour. The integrative model we offer has implications that can advance theories of cognition, enculturation, adaptation, and psychopathology. Crucially, this formal (variational) treatment seeks to resolve key debates in current cognitive science, such as the distinction between internalist and externalist accounts of theory of mind abilities and the more fundamental distinction between dynamical and representational accounts of enactivism.

Recognition

Recognition has both a normative and a psychologicaldimension. Arguably, if you recognize another person with regard to acertain feature, as an autonomous agent, for example, you do not onlyadmit that she has this feature but you embrace a positive attitudetowards her for having this feature. Such recognition implies that youbear obligations to treat her in a certain way, that is, you recognizea specific normative status of the other person, e.g., as a free andequal person. But recognition does not only matter normatively. It isalso of psychological importance. Most theories of recognition assumethat in order to develop a practical identity, persons fundamentallydepend on the feedback of other subjects (and of society as awhole). According to this view, those who fail to experience adequaterecognition, i.e., those who are depicted by the surrounding others orthe societal norms and values in a one-sided or negative way, willfind it much harder to embrace themselves and their projects asvaluable. Misrecognition thereby hinders or destroys persons’successful relationship to their selves. It has been poignantlydescribed how the victims of racism and colonialism have sufferedsevere psychological harm by being demeaned as inferior humans (Fanon1952). Thus, recognition constitutes a “vital human need”(Taylor 1992, 26).