AI Perspectives in Education: A BERT-based Exploration of Informatics Students’ Attitudes to ChatGPT

This study examined informatics students’ perspectives on ChatGPT through sentiment analysis, utilizing advanced BERT neural networks within Python to understand their sentiments comprehensively. By harnessing BERT’s intricate contextual embeddings and attention mechanisms, we effectively evaluated the subtleties of students’ attitudes. Our methodology encompassed categorizing student ChatGPT reviews using a 5-star rating scale, capturing diverse sentiments and enabling scrutiny of their distribution and trends. Positive sentiments were predominant, as students lauded ChatGPT’s potential as a valuable academic tool. Its capacity to assist in research, assignments, and information retrieval garnered significant praise. However, neutral and negative sentiments pinpointed areas for improvement. Neutral sentiments indicated potential optimization, while negative sentiments flagged inaccuracies and response time concerns. Ultimately, this amalgamation of sentiment analysis and BERT insights provided a lucid view of informatics students’ interactions with ChatGPT. In conclusion, the fusion of sentiment analysis and BERT insights has spotlighted ChatGPT’s immense potential as an academic companion. This study underscores its value as an effective support tool, emphasizing the commitment to continuous enhancement. By diligently refining ChatGPT, we can enhance its performance, tailor it more precisely to students’ diverse needs, and seamlessly empower them to integrate AI-driven language models into their educational journeys.