This study investigates public attitudes toward green hydrogen (GH) by analyzing YouTube comments through sentiment analysis and topic modelling. Unlike previous research that situates hydrogen within broader climate or energy debates and focuses on platforms such as Twitter or Bilibili, this work examines GH as a standalone topic and leverages YouTube's longer, context-rich comments to capture richer public discourse. Comments were collected via the YouTube API (Application Programming Interface) from a curated set of videos and analyzed for sentiment using both the rule-based VADER (Valence Aware Dictionary and sEntiment Reasoner) and the generative model ChatGPT-3.5, enabling a qualitative comparison of their performance. Latent Dirichlet Allocation (LDA) was then applied to identify major discussion themes, which were subsequently linked to sentiment trends. The results indicate that ChatGPT-3.5 outperforms VADER in interpreting sarcasm, slang, emoticons, and mixed sentiments. Topic modelling revealed eight key themes, including skepticism about institutional barriers and costs, optimism regarding GH's role in hard-to-decarbonize sectors, comparisons with nuclear energy and electric vehicles, and concerns about environmental and technical challenges. Overall, the study enhances understanding of online public discourse on GH by demonstrating how advanced sentiment analysis tools, combined with topic modelling, can generate deeper insights to inform strategies that better integrate public perceptions with the economic and policy conditions of GH deployment.