Selecting JavaScript packages is difficult because developers must search large repositories and judge many possible alternatives. This work compares ChatGPT with AIDT, a recommender system that combines meta-search and machine-learning techniques to rank packages for a developer query. A user study evaluates the recommendations produced by both approaches against developers’ criteria and expectations. The findings show that ChatGPT offers flexibility and can provide useful explanations for its choices, but it does not yet consistently match the effectiveness of AIDT or human selection for the task. The study highlights both the potential and the current limitations of conversational language models as assistants for software-technology selection.