Benchmarking Multimodal Large Language Models for Scientific Visualization Literacy
Artificial IntelligenceComputation and LanguageHuman-Computer Interaction
Summary
The authors tested six different multimodal large language models (MLLMs) to see how well they understand scientific visualizations, like graphs and illustrations that show scientific data. They used a special test made for checking science visualization skills and compared model results to human performance. They found that the best model, Gemini, does better than the average human, but many models still struggle, especially with detailed numbers and complex visuals. The authors suggest that understanding scientific visualizations should be an important way to check how good these AI models are.
Authors
Patrick Phuoc Do, Chau M. Ta, Chaoli Wang
Abstract
Multimodal large language models (MLLMs) are increasingly used to interpret visualizations, yet current evaluations remain largely chart-centric and provide limited evidence of understanding of scientific visualization (SciVis). We benchmark six MLLMs on the scientific visualization literacy assessment test, a standardized SciVis literacy assessment comprising 49 items based on 18 scientific visualizations and illustrations, spanning 8 techniques and 11 task types. We evaluate three closed-source and three open-source models under a closed-world protocol and compare their performance using data from 485 human participants. Results show that current MLLMs do not exhibit uniform SciVis literacy. Gemini is the strongest model overall, exceeding the human mean across the evaluated subsets, whereas the open-source models remain below the human baseline. Performance is highly uneven across techniques and tasks: models perform best on scientific illustration, search, and spatial understanding, but struggle on texture-based and integration-based visualizations and on quantitative estimation. Error analysis reveals recurring failures in fine-grained quantitative estimation, flow-direction interpretation, and grounded encoding interpretation. These findings position SciVis literacy as a necessary benchmark dimension for evaluating multimodal AI systems. Our code and model outputs are publicly available at https://github.com/patdmp/mllm-scivis-lit-benchmark.