ANÁLISE COMPARATIVA DE INTELIGÊNCIAS ARTIFICIAIS PARA PESQUISA E DESENVOLVIMENTO DE HISTÓRIAS E PERSONAGENS BASEADAS NO FOLCLORE BRASILEIRO

This research explores the potential of Brazilian folklore and literature as a source for audiovisual productions with global reach. Given the growing integration of generative artificial intelligence (AI) in academic and creative research, this study conducted a comparative analysis of four large language models (ChatGPT-4, Grok-3, DeepSeek v3, and Gemini 2.5). Methodologically, this is an applied, exploratory study with a qualitative approach, based on the collection and triangulation of data artificially generated through a standardized prompt. The results revealed that while the models consistently suggested popular figures such as Saci-Pererê and Iara, a prominent algorithmic bias toward Hollywood commercial patterns and cultural homogenization emerged. DeepSeek v3 showed greater ethnic-cultural diversity by including Indigenous and African myths , but all tools failed to access niche legends or oral traditions. It is concluded that, although AIs are valuable tools for complementing traditional folkloric research, they cannot replace it. In-depth human research remains essential to uncover the nuances and the vast range of lesser-known figures in Brazilian folklore, ensuring the authenticity and originality necessary for cultural and commercial success.