In GPT we trust : perceptions of the future of work with artificial intelligence on online forums

Abstract

This study analyzes over 10 years of Reddit discussions on artificial intelligence (AI)’s impact on labor market using a state-of-the-art pre-trained robust optimized bidirectional encoder representations from transformers (BERT) model, or RoBERTa model, for sentiment analysis and emotional detection. Additionally, the BERTopic method is employed to identify the discussed topics. The results show generally mixed sentiments towards AI’s impact on the labor market. Fueled by influencers and media sensationalism, negative emotions—fear, anger, and disgust—stem from concerns about job replacement, economic instability, and societal change. However, the release of chat generative pre-trained transformer’s (ChatGPT) in 2022 has created a more positive outlook, with users expressing surprise and optimism about generative AI’s capabilities. Discussions have shifted from fears of complete automation to the potential for augmentation. Emotional dynamics have been shaped by events such as the COVID-19 pandemic and the rise of generative AI, underscoring AI’s dual role as both a threat and an opportunity.