From Hashtag to Image: A Discursive Analysis of TikTok Trends in AI-Generated Images

Authors

DOI:

https://doi.org/10.5965/1982615x19472026200

Keywords:

fashion, artificial intelligence, representations

Abstract

This article investigates how the Generative Artificial Intelligence (GAI) model ChatGPT-4 interprets and reproduces visual discourses of femininity based on popular trends from the TikTok platform. The study aims to analyze the meaning effects present in images produced by the DALL-E model, derived from the hashtags #cleangirl and #patynordestina, observing how global and local aesthetic patterns are represented and re-signified. The research adopts a qualitative approach, grounded in bibliographic review and discursive semiotic analysis, to examine the planes of expression and content within the generated images. The results indicate that GAI tends to reproduce hegemonic patterns of beauty and behavior, reinforcing Eurocentric and idealized stereotypes. While the #cleangirl trend emphasizes the valorization of cleanliness and aesthetic neutrality, #patynordestina demonstrates the fetishization of cultural otherness, transforming regional references into stylized and superficial elements. It is concluded that GAI still lacks cultural sensitivity and diversity within its datasets, perpetuating biases and imagetic standardizations that constrain the plurality of female representations in the digital sphere.

Downloads

Download data is not yet available.

Author Biographies

Laís Fidelis Gonçalves, Universidade de São Paulo

Master’s student in Textile and Fashion at the School of Arts, Sciences and Humanities of the University of São Paulo (USP). Holds a Bachelor’s degree in Textile and Fashion from EACH – USP (2022). Member of the MODATA research group (CNPq). Research interests include FASHION DEVELOPMENT, DATA TECHNOLOGIES, ARTIFICIAL INTELLIGENCE, SUSTAINABILITY, and SEMIOTICS, with an emphasis on GENERATIVE IMAGES and DISCURSIVE SEMIOTIC ANALYSIS. Has professional experience in the STYLE department at PACE (2020) and Riachuelo (2023) and is currently a teaching assistant in the undergraduate program in Textile and Fashion at USP.

Sarah Suyama, Universidade de São Paulo

Master’s student in Textile and Fashion in the Graduate Program in Textile and Fashion at the University of São Paulo (2024). Holds a Bachelor’s degree in Textile and Fashion from the School of Arts, Sciences and Humanities of the University of São Paulo (2019), where she served as a student representative for the Textile and Fashion program for three years (2015–2018). Since 2018, she has been working in the product management field for major brands, in both domestic and international purchasing, with experience in raw materials as well as finished products.

Suzana Helena de Avelar Gomes, Universidade de São Paulo

She holds a Bachelor's degree in Fashion Design from Faculdade Santa Marcelina (1995), a Master's degree in Communication and Semiotics from the Pontifical Catholic University of São Paulo (2000), and a Ph.D. in Communication and Semiotics from the same institution (2005). She has held teaching appointments in undergraduate Fashion programs at institutions such as Faculdade Santa Marcelina, Belas Artes, UniFMU, and Universidade Anhembi Morumbi. She is currently a tenured Associate Professor at the University of São Paulo, with research and teaching interests primarily focused on the following areas: fashion, globalization, emerging technologies, digital culture, and the culture of datafication.

Cláudia Regina Garcia Vicentini, Universidade de São Paulo

Associate Professor at the School of Arts, Sciences and Humanities at the University of São Paulo (USP). Faculty member in the undergraduate and Graduate Program in Textile and Fashion. I currently serve as Vice-President of the Research and Innovation Committee at EACH and as a member of the Human Rights Committee at the same unit. I am part of the MODATA research group – Research in Fashion and Datafication. At present, my work focuses on studies that seek to address subjects and their ways of life in contemporary society; thus, I have directed my research towards the intersections of communication, image, gender, biopolitics, necropolitics, and decoloniality, from a political-philosophical perspective.

References

AIRES, Tatiana Filipa Costa. TikTok: escolhemos o que vemos ou o algoritmo escolhe por nós?. The Trends Hub, [S. l.],v. 1, n. 4, 1-8, 2024. Disponível em: https://doi.org/10.34630/tth.vi4.5701. Acesso em: 15 maio 2025.

BARROS, Diana Luz Pessoa de. Teoria semiótica do texto. 4. ed. São Paulo: Ática, 2005. 92 p.

CALEFATO, Patrizia. The clothed body. Tradução de Lisa Adams. New York: Berg, 2004. 167 p.

DE OLIVEIRA ARAÚJO, Fabrícia Raquel; PONTES, Verônica Maria de Araújo. Para além das trends: a influência do Tik Tok na formação leitora. Revista Semiárido De Visu, [S. l.], v. 13, n. 3, p. 425–435, 2025. DOI: 10.31416/rsdv.v13i3.1334. Disponível em: https://semiaridodevisu.ifsertaope.edu.br/index.php/rsdv/article/view/1334. Acesso em: 20 out. 2025.

FERRARA, Emilio. Should ChatGPT be biased? Challenges and risks of bias in large language models. First Monday, v. 28, n. 11. Novembro, 2023. Disponível em: https://firstmonday.org/ojs/index.php/fm/article/view/13346. Acesso em: 19 out. 2025.

FIORIN, José Luiz. Elementos de análise do discurso. 9 ed. São Paulo: Contexto, 2000. 93 p.

FIORIN, José Luiz. Linguagem e ideologia. 1 ed. São Paulo: Ática, 2011.

GREGOLIN, Maria do Rosario Valencise. A análise do discurso: conceitos e aplicações. Alfa: Revista de Linguística, São Paulo, v. 39, 2001. p. 13-21. Disponível em: https://periodicos.fclar.unesp.br/alfa/article/view/3967. Acesso em: 15 maio 2025.

GREIMAS, Algirdas Julien. Semiótica figurativa e semiótica plástica. Significação: Revista Brasileira de Semiótica, n. 4, jun. 1984. p. 18-187.

LIPOVETSKY, G. O império do efêmero. São Paulo: Companhia das Letras, 1989.

LUDERMIR, Teresa Bernarda. Inteligência Artificial e Aprendizado de Máquina: estado atual e tendências. Estudos Avançados, v. 35, n. 101, São Paulo. Abril, 2021. p. 37-50. Disponível em: ht-tps://doi.org/10.1590/s0103-4014.2021.35101.007. Acesso em: 19 out. 2025.

MCKINSEY & COMPANY. The state of AI in early 2024: gen AI adoption spikes and starts to generate value. McKinsey & Company, 2024. Disponível em: https://www.mckinsey.com. Acesso em: 15 maio 2025.

NEGRI, Antoni. Biocapitalismo. São Paulo: Iluminuras, 2015. Versão Kindle.

OPENAI. Hello GPT-4o. OpenAI, maio 2024. Disponível em: https://openai.com/index/hello-gpt-4o/. Acesso em: 15 maio 2025.

OPENAI. ChatGPT — Release Notes. [S. l.], 10 abr. 2025. Dis-ponível em: https://help.openai.com/en/articles/6825453-chatgp-t-release-notes. Acesso em: 19 out. 2025.

ORLANDI, Eni Puccinelli. Análise de discurso: princípios & procedimentos. 8. ed. Campinas: Pontes, 2009. p. 9-100.

ROSSI, Túlio Cunha. Feminilidade e suas imagens em mídias digitais: questões para pensar gênero e visualidade no sé-culo XXI. Tempo Social, São Paulo, v. 29, n. 1, p. 235-255, 2017.

SICHMAN, Jaime. Inteligência Artificial e sociedade: avanços e riscos. Estudos Avançados, v. 35, n. 101, São Paulo, jan-abr. 2021. p. 37-50. Disponível em: https://doi.org/10.1590/s0103-4014.2021.35101.004. Acesso em: 15 maio 2025.

SILVA, Larissa et al. Indexação social como estratégia para visibilidade de conteúdos científicos em serviço de rede social online TikTok. IN: TOGNOLI, Natália Bolfarini; ALBUQUERQUE, Ana Cristina de; CERVANTES, Brígida Maria Nogueira (orgs.). Organização e representação do conhecimento em diferentes contextos: desafios e perspectivas na era da datificação. Londrina, ISKO-Brasil; PPGCI-UEL, v. 6, 2023. p. 335-346. Disponível em: https://livroaberto.ufpa.br/items/303565eb-1870-4d76-98fe-d426b0771734. Acesso em: 15 maio 2025.

SILVEIRA, Sérgio Amadeu da. Democracia e os códigos invisíveis: como os algoritmos estão modulando comportamentos e escolhas políticas. São Paulo: Sesc SP, 2019.

SOUZA, Tania C. Clemente de. Discurso e imagem: perspectivas de análise não verbal. Ciberlegenda, n. 1, 1998. Disponível em: https://periodicos.uff.br/ciberlegenda/article/view/36741

TIKTOK FOR BUSINESS. TikTok + WGSN: Como as comunidades vão impactar as tendências no digital?. TikTok For Business, jul. 2023. Disponível em: https://ads.tiktok.com/business/pt-BR/blog/wgsn-comportamento-tendencia-digital-comunidades. Acesso em: 15 maio 2025.

Downloads

Published

2026-01-01

How to Cite

GONÇALVES, Laís Fidelis; SUYAMA, Sarah; GOMES, Suzana Helena de Avelar; VICENTINI, Cláudia Regina Garcia. From Hashtag to Image: A Discursive Analysis of TikTok Trends in AI-Generated Images. ModaPalavra e-periódico, Florianópolis, v. 19, n. 47, p. 200–230, 2026. DOI: 10.5965/1982615x19472026200. Disponível em: https://www.periodicos.udesc.br/index.php/modapalavra/article/view/27617. Acesso em: 9 jan. 2026.

Issue

Vol. 19 No. 47 (2026): Technofashion: the techniques that weave and are woven by fashion (V. 19, N. 47)

Section

Dossier

License

Copyright (c) 2026 Laís Fidelis Gonçalves, Sarah Suyama, Suzana Helena de Avelar Gomes, Cláudia Regina Garcia Vicentini

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

When submitting an article for publication in ModaPalavra e-periodico, the author (s) agree (s) with the following terms:

  1. Authors maintain the copyright in their manuscripts and grant the journal the right of first publication, with work simultaneously licensed under the Creative Commons Attribution-NonCommercial 4.0 International, which allows sharing the work with the acknowledgment of authorship and the initial publication in the journal without payment ;
  2. Authors may use the same results in other publications after the first publication, provided that they indicate ModaPalavra e-journal as the original publication medium;
  3. Authors are authorized to take additional contracts, separately, only after the original publication in ModaPalavra e-journal, provided they indicate ModaPalavra e-journal as the original publication medium;
  4. Authors are allowed and encouraged to publish and distribute their work online (eg in institutional repositories or on their personal page), only after the editorial process and the first publication, provided they indicate ModaPalavra e-journal as the original publication medium;
  5. To indicate ModaPalavra e-journal as the original publication medium, authors should use the following text template: "This article was originally published by ModaPalavra e-periodical, under a CC BY NC license, in its volume [insert volume] number [insert number] in the year of [insert year], and can be accessed at: http://www.revistas.udesc.br/index.php/modapalavra/ ";
  6. The opinions expressed in the articles are the author’s sole responsibility, not necessarily reflecting the journal’s opinion. The publication of any material that is owned and held in copyright by a third party, including – but not limited to - articles, photos or drawings was previously authorized by their representatives to be published in ModaPalavra e-journal.
Crossref
Scopus
Google Scholar
Europe PMC

Most read articles by the same author(s)