Pubblicazioni

Autori:

Bombieri, Marco; Rospocher, Marco

Titolo:

Mining Impersonification Bias in LLMs via Survey Filling

Anno:

2025

Tipologia prodotto:

Articolo in Rivista

Tipologia ANVUR:

Articolo su rivista

Lingua:

Inglese

Referee:

Sì

Nome rivista:

INFORMATION

ISSN Rivista:

2078-2489

N° Volume:

16

Numero o Fascicolo:

11

Intervallo pagine:

1-21

Parole chiave:

large language models, personas, bias, stereotypes

Breve descrizione dei contenuti:

In this paper, we introduce a survey-based methodology to audit LLM-generated personas by simulating 200 US residents and collecting responses to socio-demographic questions in a zero-shot setting. We investigate whether LLMs default to standardized profiles, how these profiles differ across models, and how conditioning on specific attributes affects the resulting portrayals. Our findings reveal that LLMs often produce homogenized personas that underrepresent demographic diversity and that conditioning on attributes such as gender, ethnicity, or disability may trigger stereotypical shifts. These results highlight implicit biases in LLMs and underscore the need for systematic approaches to evaluate and mitigate fairness risks in model outputs.

Pagina Web:

https://www.mdpi.com/2078-2489/16/11/931

Id prodotto:

148137

Handle IRIS:

11562/1174028

ultima modifica:

19 dicembre 2025

Citazione bibliografica:

Bombieri, Marco; Rospocher, Marco, Mining Impersonification Bias in LLMs via Survey Filling «INFORMATION» , vol. 16 , n. 11 , 2025 , pp. 1-21

Consulta la scheda completa presente nel repository istituzionale della Ricerca di Ateneo