Research on AI has extensively considered biases related to gender and race. However, much less attention has been dedicated to another sociological tenet: that of class. Inspired by Bourdieu’s work on cultural stratification and distinction, this work sheds light on the sociocultural roots of artificial sociality, and on how these become manifest as ‘habitus’ within the outputs of generative AI models. We conducted 39 interviews with three AI chatbots – ChatGPT, Gemini and Replika – after asking them to impersonate individuals with different occupational positions: highly skilled professionals, blue-collar workers, university professors in the humanities, construction workers, computer scientists and hairdressers. Our qualitative study shows class-based regularities in how popular AI chatbots represent the lifestyle and tastes of fictional personas in artificial conversations, partly mediated by infrastructural and design elements. The article proposes a sociological perspective on bias in artificial sociality and experiments with interview methods in the study of generative AI.
AiroldiM (2022) Machine Habitus: Toward a Sociology of Algorithms. Cambridge: Polity Press.
2.
Arseniev-KoehlerAFosterJG (2022) Machine learning as a model for cultural learning: teaching an algorithm what it means to be fat. Sociological Methods & Research51(4): 1484–1539.
3.
BachmayerTWilterdinkNvan VenrooijA (2014) Taste differentiation and hierarchization within popular culture: the case of salsa music. Poetics47: 60–82.
4.
BalmerA (2023) A sociological conversation with ChatGPT about AI ethics, affect and reflexivity. Sociology57(5): 1249–1258.
5.
BenderEMGebruTMcMillan-MajorA, et al. (2021) On the dangers of stochastic parrots: can language models be too big? In: Proceedings of the 2021 ACM conference on fairness, accountability, and transparency (FAccT’21), New York, USA, 1 March, pp. 610–623. New York: Association for Computing Machinery.
6.
BourdieuP (1984) Distinction: A Social Critique of the Judgement of Taste. Cambridge, MA: Harvard University Press.
7.
BourdieuP (1989) Social space and symbolic power. Sociological Theory7(1): 14–25.
8.
BourdieuP (1990) The Logic of Practice. Stanford, CA: Stanford University Press.
9.
BourdieuP (1991) Language and Symbolic Power. Cambridge, MA: Harvard University Press.
10.
BurkhardtSRiederB (2024) Foundation models are platform models: prompting and the political economy of AI. Big Data & Society11: 1–15.
11.
CaliskanABrysonJJNarayananA (2017) Semantics derived automatically from language corpora contain human-like biases. Science356(6334): 183–186.
12.
ChanTWGoldthorpeJH (2007) Class and status: the conceptual distinction and its empirical relevance. American Sociological Review72(4): 512–532.
13.
ChildressCBaumannSRawlingsCM, et al. (2021) Genres, objects, and the contemporary expression of higher-status tastes. Sociological Science8: 230–264.
14.
CotterK (2019) Playing the visibility game: how digital influencers and algorithms negotiate influence on Instagram. New Media & Society21(4): 895–913.
15.
de SetaGPohjonenMKnuutilaA (2023) Synthetic ethnography: field devices for the qualitative study of generative models. SocArXiv. Available at: https://osf.io/zvew4
16.
DengelAGehrleinRFernesD, et al. (2023) Qualitative research methods for large language models: conducting semi-structured interviews with ChatGPT and BARD on computer science education. Informatics10(4): 78.
17.
DepountiISaukkoPNataleS (2023) Ideal technologies, ideal women: AI and gender imaginaries in Redditors’ discussions on the Replika bot girlfriend. Media, Culture & Society45(4): 720–736.
18.
DieterMGerlitzCHelmondA, et al. (2019) Multi-situated app studies: methods and propositions. Social Media + Society5(2): 1–15.
19.
EspositoE (2022) Artificial Communication: How Algorithms Produce Social Intelligence. Strong Ideas Series. Cambridge, MA: The MIT Press.
20.
EubanksV (2018) Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor. New York City: St. Martin’s Press.
21.
FerraraE (2023) Should ChatGPT be biased? challenges and risks of bias in large language models. First Monday28(11).
22.
GoffmanE (1959) The Presentation of Self in Everyday Life. London: Penguin.
LightBBurgessJDuguayS (2018) The walkthrough method: an approach to the study of apps. New Media & Society20(3): 881–900.
27.
LizardoO (2016) Why ‘cultural matters’ matter: culture talk as the mobilization of cultural capital in interaction. Poetics58: 1–17.
28.
McGuffieKNewhouseA (2020) The radicalization risks of GPT-3 and advanced neural language models. arXiv. Available at: http://arxiv.org/abs/2009.06807
29.
MageeLAroraVMunnL (2023) Structured like a language model: analysing AI as an automated subject. Big Data & Society10: 1–20.
30.
MiceliMPosadaJYangT (2022) Studying up machine learning data: why talk about bias when we mean power?Proceedings of the ACM on Human-Computer Interaction6: 1–34.
MotokiFPinho NetoVRodriguesV (2024) More human than human: measuring ChatGPT political bias. Public Choice198(1): 3–23.
33.
MühlhoffR (2020) Human-aided artificial intelligence: or, how to run large computations in human brains? toward a media sociology of machine learning. New Media & Society22(10): 1868–1884.
34.
NadeemAMarjanovicOAbedinB (2022) Gender bias in AI-based decision-making systems: a systematic literature review. Australasian Journal of Information Systems26: 1–34.
PaßmannJSchubertC (2021) Liking as taste making: social media practices as generators of aesthetic valuation and distinction. New Media & Society23(10): 2947–2963.
39.
PentinaIHancockTXieT (2023) Exploring relationship development with social chatbots: a mixed-method study of Replika. Computers in Human Behavior140: 107600.
40.
PerrottaCSelwynNEwinC (2022) Artificial intelligence and the affective labour of understanding: the intimate moderation of a language model. New Media & Society26(3): 1585–1609.
SavageMDevineFCunninghamN, et al. (2013) A new model of social class? findings from the BBC’s great British class survey experiment. Sociology47(2): 219–250.
45.
SkjuveMFølstadAFostervoldKI, et al. (2022) A longitudinal study of human–chatbot relationships. International Journal of Human-Computer Studies168: 102903.
46.
SweeneyM (2016) The Ms. Dewey ‘experience’: technoculture, gender, and race. In: DanielsJGregoryKCottomTM (eds) Digital Sociologies. Bristol: Policy Press, pp. 401–420.
47.
ThakurV (2023) Unveiling gender bias in terms of profession across LLMs: analyzing and addressing sociological implications. arXiv:2307.09162. arXiv. Available at: http://arxiv.org/abs/2307.09162 (accessed 29 March 2024).
WebsterJ (2023) The promise of personalisation: exploring how music streaming platforms are shaping the performance of class identities and distinction. New Media & Society25(8): 2140–2162.
WoolgarS (1985) Why not a sociology of machines? the case of sociology and artificial intelligence. Sociology19(4): 557–572.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
