IA e algoritmos enviesados: as dimensões do viver na sociedade 5.0

• Autor(a) principal: Samarão, Liliany Alves
• Data de Publicação: 2023
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da PUC_SP
• Texto Completo: https://repositorio.pucsp.br/jspui/handle/handle/40758
• topic: CNPQ::CIENCIAS SOCIAIS APLICADAS::COMUNICACAO; Sociedade 5.0; Inteligência artificial; Racismo algorítmico; Algoritmos enviesados; Corpomídia; Society 5.0; Artificial intelligence; Algorithmic racism; Biased algorithms; Corpusmedia

The world was still adapting to the advances of Industry 4.0 – which radically transformed societies – when Society 5.0 emerged. In it, the social organizational model becomes mainly based on technology (Big Data, Artificial Intelligence, Internet of Things) presented assolution to all human needs. Society 5.0 is connected by an avalanche of data that circulates online and is responsible for feeding technology. Artificial Intelligence talks to the individual, offers them products, answers their questions, negotiates their debts, produces medical diagnoses, helps the police identify possible criminals, writes texts and makes art. Intellectual, emotional, social and cultural life started to be transformed into gadgets, turning the individual into a special commodity. However, technology is not neutral. What we don't see, in the process, are the human decisions that were made during the development of these technologies an how the developers assumptions, values, beliefs shapes them. Hence, this research begins with the hypothesis that it is essential to comprehend AI and its "learning" organization by examining the social dimension and striving to identify the biased tendencies of machines that generate biased algorithms, which have serious technological and social consequences. The term "biased algorithms" has been employed with the aim of further expanding this important discussion. The objective of this research is to demonstrate that algorithms are not neutral and may exhibit biases. Biases represent acquired assumptions, beliefs, or attitudes that have the potential to perpetuate stereotypes. Biases can encompass prejudices related to gender, race, age, or appearance. Bias may be rooted in stereotypes,subjective judgments, and pre-existing prejudices. Therefore, we underpin our work with the concept of "biased algorithms." And as all technology changes the body-environment relationship that guides life, this research takes as its object the investigation of new AI tools, their sometimes biased results and how they are transforming societies - societies connected with increasingly complex systems and that move between the online and offline worlds, releasing a set of data, files and documents that become available for unknown uses. Today, it has become urgent to understand how AI operates in this new model of society – how it will increasingly change health, transportation, finance and manufacturing. How it will become essential in diagnostics, autonomous vehicles, smart manufacturing and personalization of education – which will see significant advances in the coming years. Therefore, the goal is to investigate the new 5.0 society and then understand the ethical issues such as transparency, privacy, accountability and algorithmic biases that have become more urgent as AI becomes an even more integrated parte of everyday life. The ongoing debate about how to properly regulate and control AI is crucial. Can the individual position themselves as an active subject in the construction of this new society increasingly guided by AI? Are we automating prejudice and racism in our increasingly connected society? What are the social implications of AI? In its theoretical foundation, there is the Teoria Corpomidia (Katz and Greiner), studies and researches from the largest universities in the world, such as MIT and Princeton, brings compilations from major journals, such as Technology Review and works from institutes focused on the expansion of AI, such as the Center for AI Safety and the Council of Europe, in addition to several authors focused on the topic of AI such as Kearns and Roth; Wachter-Boettcher and Joy Buolamwini