La inteligencia artificial para el diseño de personajes en videojuegos: estudio de sesgos y estereotipos en Midjourney®
Enviado: 05-02-2025
|Aceptado: 26-02-2025
|Publicado: 31-03-2025
Derechos de autor 2025 Javier Corzo Martínez, Manuel Drago Díaz Alemán, Jorge de La Torre Cantero
Esta obra está bajo una licencia internacional Creative Commons Atribución-NoComercial-CompartirIgual 4.0.
Descargas
Palabras clave:
Inteligencia artificial, concept art, sesgos, estereotipos, Midjourney
Agencias de apoyo:
Resumen:
Las inteligencias artificiales generativas de imágenes han impactado el arte de los videojuegos y a la industria audiovisual, modificando el trabajo de ilustradores y concept artists. Numerosos estudios han demostrado que estas herramientas reproducen estereotipos de edad, género y etnia, lo que plantea preocupaciones éticas. Este estudio analizó Midjourney®, evaluando las imágenes generadas a partir de cinco categorías de prompts. Los resultados revelaron sesgos estructurales y fallos en los filtros NSFW (not safe for work), que en ocasiones produjeron resultados opuestos a su propia finalidad. Dado que Midjourney® refuerza estereotipos sociales y culturales, se enfatiza la necesidad de supervisión humana en la creación de personajes y se invita a que los desarrolladores de este tipo de software atiendan a las inequidades que sus productos ayudan a reforzar.
Citas:
Alenichev, A., Kingori, P. y Grietens, K. P. (2023). Reflections before the storm: the AI reproduction of biased imagery in global health Visuals. The Lancet Global Health, 11(10), e1496-e1498. https://doi.org/10.1016/S2214-109X(23)00329-7
Barocas, S., Hardt, M. y Narayanan, A. (2023). Fairness and Machine learning. Limitations and opportunities. Introduction. Fairmlbook.org. MIT Press. https://fairmlbook.org/introduction.html
Bianchi, F., Kalluri, P., Durmus, E., Ladhak, F., Cheng, M., Nozza, D., Hashimoto, T., Jurafsky, D., Zou, J. y Caliskan, A. (2023). Easily accessible text-to-image generation amplifies demographic stereotypes at large scale. Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency, 1493-1504. https://doi.org/10.1145/3593013.3594095
Birhane, A., Prabhu, V. U. y Kahembwe, E. (2021). Multimodal datasets: misogyny, pornography, and malignant stereotypes. arXiv preprint arXiv:2110.01963. https://doi.org/10.48550/arXiv.2110.01963
Common Crawl. (2023). Open Repository of Web Crawl Data. https://commoncrawl.org/
Davis, W. (2023). Valve won’t approve Steam games that use copyright-infringing AI artwork. The Verge. https://www.theverge.com/2023/7/1/23781339/valve-steam-ai-artwork-rejecting-banning-pc-games
Dhariwal, P. y Nichol, A. (2021). Diffusion models beat GANs on image synthesis. Advances in Neural Information Processing Systems, 34, 8780-8794. https://doi.org/10.48550/arXiv.2105.05233
Dunlap, M. (2023). StableDiffusion2-Image-to-Text. GitHub. https://github.com/mddunlap924/StableDiffusion2-Image-to-Text
Dwyer, B. y Brems, M. (1 de septiembre de 2024). What is CLIP? Contrastive Language-Image Pre-Processing Explained. Roboflow. https://blog.roboflow.com/openai-clip/
Elastic.co (s. f.). What is a large language model (LLM)? Elastic.co. https://www.elastic.co/what-is/large-language-models
Elgar, M. (2018). Are redheads with blue eyes really going extinct? Find an expert. University of Melbourne. https://findanexpert.unimelb.edu.au/news/1167-are-redheads-with-blue-eyes-really-going-extinct%3F
Friedman, B. y Nissenbaum, H. (1996). Bias in computer systems. ACM Transactions on Information Systems (TOIS), 14(3), 330-347. https://dl.acm.org/doi/10.1145/230538.230561
Frolov, S., Hinz, T., Raue, F., Hees, J. y Dengel, A. (2021). Adversarial text-to-image synthesis: A review. Neural Networks, 144, 187-209. https://doi.org/10.1016/j.neunet.2021.07.019
Guiness, H. (18 de septiembre de 2024). The Best AI Image Generators in 2024. Zapier. https://zapier.com/blog/best-ai-image-generator/
Heuer, C. A., McClure, K. J. y Puhl, R. M. (2011). Obesity stigma in online news: a visual content analysis. Journal of Health Communication, 16(9), 976-987. https://doi.org/10.1080/10810730.2011.561915
Hoffmann, A. L. (30 de abril de 2018). Data Violence and How Bad Engineering Choices Can Damage Society. Medium. https://medium.com/@annaeveryday/data-violence-and-how-bad-engineering-choices-can-damage-society-39e44150e1d4
Jiang, H. H., Brown, L., Cheng, J., Khan, M., Gupta, A., Workman, D., Hanna, A., Flowers, J. y Gebru, T. (2023). AI Art and its Impact on Artists. Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society, 363-374. https://doi.org/10.1145/3600211.3604681
King, M. (2022). Harmful biases in artificial intelligence. The Lancet Psychiatry, 9(11), e48. https://doi.org/10.1016/S2215-0366(22)00312-1
Lanz, J. M. (5 de septiembre 2023). Epic Games Embraces AI In Games, In Break with Steam. Decrypt.co. https://decrypt.co/155023/epic-games-embraces-ai-in-games-in-break-with-steam
Lin, T. Y., Maire, M., Belongie, Bourdev, L., Girshick, R., S., Hays, J., Perona, P., Ramanan, D., Zitnick, C. L. y Dollár, P. (2014). Microsoft COCO: Common Objects in Context. Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13, 740-755. Springer International Publishing. https://doi.org/10.48550/arXiv.1405.0312
Luccioni, A. S. y Viviano, J. D. (2021). What's in the Box? A Preliminary Analysis of Undesirable Content in the Common Crawl Corpus. arXiv preprint arXiv:2105.02732. https://doi.org/10.48550/arXiv.2105.02732
Mendelovich, Y. (15 de febrero de 2023). Midjourney Is Being Class-Action Sued for Severe Copyright Infringements. Y. M. Cinema Magazine. https://ymcinema.com/2023/02/15/midjourney-is-being-class-action-sued-for-severe-copyright-infringements/
Midjourney (s. f). Model Version 5. Midjourney https://docs.midjourney.com/docs/en/model-version-5
Milmo, D. y Hern, A. (28 de febrero de 2024). Google chief admits “biased” AI tool’s photo diversity offended users. The Guardian. https://www.theguardian.com/technology/2024/feb/28/google-chief-ai-tools-photo-diversity-offended-users
Nguyen, D. (2023). The Effects of AI on Digital Artist [Tesis de grado]. Haaga-Helia University of Applied Sciences Business Information Technology – BITE. Finlandia. https://urn.fi/URN:NBN:fi:amk-202305088212
Nilsback, M. E. y Zisserman, A. (2008). Automated flower classification over a large number of classes. 2008 Sixth Indian conference on computer vision, graphics & image processing, 722-729. IEEE. https://doi.org/10.1109/ICVGIP.2008.47
O’Connor, R. (12 de mayo de 2022). Introduction to Diffusion Models for Machine Learning. AssemblyAI. https://www.assemblyai.com/blog/diffusion-models-for-machine-learning-introduction/
O’Connor, R. (19 de abril de 2023). How physics advanced Generative AI. AssemblyAI. https://www.assemblyai.com/blog/how-physics-advanced-generative-ai/#generative-ai-with-thermodynamics
Parrish, A. (16 de enero de 2024). Square Enix says it used AI art in upcoming Foamstars game. The Verge. https://www.theverge.com/2024/1/16/24040124/square-enix-foamstars-ai-art-midjourney
Patton, T. O. (2006). Hey girl, am I more than my hair?: African American women and their struggles with beauty, body image, and hair. NWSA Journal, 18(2), 24-51. https://www.jstor.org/stable/4317206
Porokh, A. (19 de octubre de 2023). How will AI disrupt video game industry in 2023? Kevuru Games. https://kevurugames.com/blog/how-ai-is-disrupting-the-video-game-industry/
Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G. y Sutskever, I. (2021). Learning transferable visual models from natural language supervision. International conference on machine learning, 8748-8763. PMLR. https://doi.org/10.48550/arXiv.2103.00020
Rashtchian, C., Young, P., Hodosh, M. y Hockenmaier, J. (2010). Collecting Image Annotations Using Amazon's Mechanical Turk. Proceedings of the NAACL HLT 2010 Workshop on Creating Speech and Language Data with Amazon's Mechanical Turk. https://vision.cs.uiuc.edu/pascal-sentences/
Rombach, R., Blattmann, A., Lorenz, D., Esser, P. y Ommer, B. (2022). High-resolution image synthesis with latent diffusion models. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 10684-10695. https://doi.org/10.48550/arXiv.2112.10752
Said, E. W. (1985). Orientalism Reconsidered. Cultural Critique, 1, 89–107. https://doi.org/10.2307/1354282
Schuhmann, C., Beaumont, R., Vencu, R., Gordon, C., Wightman, R., Cherti, M., Coombes, T., Katta, A., Mullis, C., Wortsman, M., Schramowski, P., Kundurthy, S., Crowson, K., Schmidt, L., Kaczmaczyk, R. y Jitsev, J. (2022). Laion-5b: An open large-scale dataset for training next generation image-text models. Advances in Neural Information Processing Systems, 35, 25278-25294. https://doi.org/10.48550/arXiv.2210.08402
Schuhmann, C., Vencu, R., Beaumont, R., Kaczmarczyk, R., Mullis, C., Katta, A., Coombes, T., Jitsev, J. y Komatsuzaki, A. (2021). Laion-400m: Open dataset of clip-filtered 400 million image-text pairs. arXiv preprint arXiv:2111.02114. https://doi.org/10.48550/arXiv.2111.02114
Selbst, A. D., Boyd, D., Friedler, S. A., Venkatasubramanian, S. y Vertesi, J. (2019). Fairness and abstraction in sociotechnical systems. Proceedings of the conference on Fairness, Accountability, and Transparency, 59-68. https://doi.org/10.1145/3287560.3287598
Stable Diffusion (2023). Frequently asked questions. Stable Diffusion https://stablediffusionweb.com/
Steam. (10 de enero de 2024). AI content on Steam. Steam Community. https://steamcommunity.com/groups/steamworks/announcements/detail/3862463747997849619
The Artificial Intelligence Channel. (10 de diciembre de 2017). The Trouble with Bias - NIPS 2017 Keynote - Kate Crawford #NIPS2017 [Archivo de Vídeo]. YouTube. https://www.youtube.com/watch?v=fMym_BKWQzk
The Media Majlis (2018). Interview with Viola Shafik. [Archivo de Vídeo]. The Media Majlis at Northwestern University in Qatar. https://mediamajlis.northwestern.edu/en/explore/explore-content/2018-18-5-interview-interview-with-viola-shafik-shafik-viola-the-media-majlis-at-northwestern-university-in-qatar-2018-2018-18
Traylor. J. (27 de julio de 2022). No quick fix: How OpenAI's DALL·E 2 illustrated the challenges of bias in AI. NBC News. https://www.nbcnews.com/tech/tech-news/no-quick-fix-openais-dalle-2-illustrated-challenges-bias-ai-rcna39918
Vahdat, A. y Kreis, K. (26 de abril de 2022). Improving diffusion models as an alternative to GANs, Part 1. NVIDIA Technical Blog. https://developer.nvidia.com/blog/improving-diffusion-models-as-an-alternative-to-gans-part-1/
Wah, C., Branson, S., Welinder, P., Perona, P. y Belongie, S. (2011). Caltech-UCSD Birds-200-2011 (CUB-200-2011). California Institute of Technology. https://www.vision.caltech.edu/datasets/cub_200_2011/
Wolfe, R. y Caliskan, A. (2022). Markedness in visual semantic AI. Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, 1269-1279. https://doi.org/10.1145/3531146.3533183
Young, P., Lai, A., Hodosh, M. y Hockenmaier, J. (2014). From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptions. Transactions of the Association for Computational Linguistics, 2, 67-78. https://doi.org/10.1162/tacl_a_00166
Zhang, H. (6 de noviembre de 2023). Xbox and Inworld AI partner to empower game creators with the potential of generative AI. Microsoft Game Dev. https://developer.microsoft.com/en-us/games/articles/2023/11/xbox-and-inworld-ai-partnership-announcement/
