Curadoria algorítmica nas plataformas de streaming e os efeitos adversos dos sistemas de recomendação

• Ano de defesa: 2023
• Autor(a) principal: Lima, Lucas Peixoto de
• Orientador(a): Sá Neto, Arthur Autran Franco de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência, Tecnologia e Sociedade - PPGCTS
• Palavras-chave em Português: Sistemas de recomendação; Algoritmos inteligentes; Tecnologia; Indústria cultural
• Palavras-chave em Inglês: Plataformas de streaming
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; CIENCIAS HUMANAS::SOCIOLOGIA; CIENCIAS SOCIAIS APLICADAS::COMUNICACAO
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/19006

This research, within the scope of studies in Science, Technology and Society, has as its theme the relationship between the cultural industry, artificial intelligence and cultural diversity, in which it studies if, and how, the recommendation systems of the streaming platforms Globoplay and Netflix act towards the creation of social bubbles due to the personalization, made through the consumption of audiovisual materials. It is justified in the sense of exploring the effects caused by intelligent algorithms, whether intentional or not, which go unnoticed by the public and can influence their ways of thinking and seeing the world. The objective, therefore, is to understand the performance of recommendation systems on audiovisual streaming platforms. The hypothesis raised is that the intelligent algorithms of recommendation systems contribute to the formation of social bubbles, in order to reaffirm users' social, cultural and political convictions. The research is divided into four fundamental stages: the description of the research objects, the survey of Latin American titles present on the platforms, the analysis of the home screen of these platforms and the discussion about algorithms and society. The characteristic of a recommendation system, based on AI and big data, of learning from the user and indicating content that matches their profile, gives evidence that the hypothesis may be true, as the user's profile, films and series that relate to this profile can be more recommended to him. However, the opacity of streaming platforms does not allow us to reach an affirmative conclusion. One solution suggested for the problems raised here focuses on transparency, allowing users to access their profile records and adjust what is recommended, and also on the regulation of platforms by the State.