Gender bias in LLMs: analyzing AI responses on STEM career guidance
Cuadernos.info · 29.05.2026 · (64) · 120-143
DOI: 10.7764/cdi.64.98546 ↗Kurz gefasst
Fachartikel in der Zeitschrift Cuadernos.info, online erschienen am 29.05.2026. Verfasst von Laura Milagros Castro Souto und 1 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-SA 4.0).
Abstract (Originalsprache)
This study examines how large language models (LLMs), specifically ChatGPT, reproduce gender bias in academic guidance toward STEM careers. Controlled interactions were conducted, varying only the gender of the young interlocutor seeking professional recommendations. The methodology included quantitative and qualitative analyses of examples, motivational language, and suggestions for external validation. The results show that, although the suggested fields of study did not differ significantly, responses to female users included more examples of traditionally feminized professions, emphasized caregiving traits, and advised validating their choices with family members or teachers. In contrast, responses to male users more frequently featured technical or leadership-oriented examples, motivation based on personal skills, and recommendations to consult professionals. These differences reflect and may perpetuate social stereotypes that contribute to the gender gap in STEM. The conclusions highlight the need for responsible AI design, improved digital literacy, and the incorporation of critical perspectives to mitigate such biases in automated tools that influence crucial educational decisions.
Zitieren
APA
Souto, L. M. C., & Castillo, G. P. (2026). Gender bias in LLMs: analyzing AI responses on STEM career guidance. Cuadernos.info, 120-143. https://doi.org/10.7764/cdi.64.98546
Vancouver
Souto LMC, Castillo GP. Gender bias in LLMs: analyzing AI responses on STEM career guidance. Cuadernos.info. 2026:120-143. doi:10.7764/cdi.64.98546
BibTeX
@article{Souto2026,
author = {Souto, Laura Milagros Castro and Castillo, Graciela Padilla},
title = {Gender bias in LLMs: analyzing AI responses on STEM career guidance},
journal = {Cuadernos.info},
year = {2026},
number = {64},
pages = {120--143},
doi = {10.7764/cdi.64.98546}
}
EUVIMED