Evaluating ChatGPT to Support Data Visualization Design

Large language models (LLMs) can help retrieve information to answer questions, construct images and audio, and assist in complex activities such as data visualization design. The latter requires specific knowledge that can be found on the internet and therefore used to train LLMs. This work investigates the ability of ChatGPT to assist in data visualization design. We conduct a metrics-based evaluation of the model and plan to expand it to understand the views of users who create visualizations, whether they are experts or not.