Análise das percepções públicas sobre a inteligência artificial em redes sociais por meio da análise de sentimentos

Abstract

In recent years, the rapid expansion of artificial intelligence (AI) has sparked intense debates about its social, ethical, and technological impacts. This study analyzes social perceptions of AI based on public discussions on Reddit, employing Natural Language Processing (NLP) techniques and sentiment analysis to identify how different communities express opinions, emotions, and stances regarding technological progress. The objective of this research is to understand how social discourse on AI has evolved over time, considering four representative historical milestones: ChatGPT (2022–2023), AI in Images (2023), DeepSeek (2024), and ChatGPT (2025). Data were collected through the Reddit API using the PRAW library in a Python environment and organized into a dataset of approximately ten thousand posts and comments in Portuguese and English. The methodological process included the stages of data collection, cleaning, text preprocessing, vectorization, and sentiment classification, using supervised machine learning models such as Naive Bayes and deep learning models such as BERT. The results indicated a predominance of positive perceptions, especially in the initial periods, followed by an increase in critical posts associated with ethical, privacy, and authorship concerns. A discursive maturation was also observed over the years, transitioning from initial enthusiasm to a more balanced and pragmatic view of AI’s role in contemporary society. Reddit proved to be a rich and representative source for sociotechnological analyses due to its diversity of communities and the authenticity of its discussions. The combination of NLP and sentiment analysis proved effective in capturing emotional and semantic nuances on a large scale, contributing to a deeper understanding of the relationship between society, language, and technology.