Figures as Interfaces: Toward LLM-Native Artifacts for Scientific Discovery

2026-04-09

AI Summary

Traditional scientific figures, as static images, lack traceability of data provenance, analysis code, and visualization standards, limiting their interpretability and manipulability by large language models (LLMs) and thereby hindering human–AI collaborative research efficiency. This work introduces the concept of “LLM-native figures,” reconceptualizing scientific visualizations as interactive, interpretable, and self-contained intelligent interfaces that embed comprehensive provenance metadata. By establishing bidirectional mappings between charts and underlying data, parsing natural language instructions, and integrating formal visualization specifications, the approach constructs a hybrid language–vision interaction framework. Validated in the domain of science of science, this method enhances research reproducibility, accelerates discovery workflows, and improves the transparency of human–AI collaborative reasoning, offering a general-purpose framework for modern scientific artifacts.

Abstract

Large language models (LLMs) are transforming scientific workflows, not only through their generative capabilities but also through their emerging ability to use tools, reason about data, and coordinate complex analytical tasks. Yet in most human-AI collaborations, the primary outputs, figures, are still treated as static visual summaries: once rendered, they are handled by both humans and multimodal LLMs as images to be re-interpreted from pixels or captions. The emergent capabilities of LLMs open an opportunity to fundamentally rethink this paradigm. In this paper, we introduce the concept of LLM-native figures: data-driven artifacts that are simultaneously human-legible and machine-addressable. Unlike traditional plots, each artifact embeds complete provenance: the data subset, analytical operations and code, and visualization specification used to generate it. As a result, an LLM can "see through" the figure--tracing selections back to their sources, generating code to extend analyses, and orchestrating new visualizations through natural-language instructions or direct manipulation. We implement this concept through a hybrid language-visual interface that integrates LLM agents with a bidirectional mapping between figures and underlying data. Using the science of science domain as a testbed, we demonstrate that LLM-native figures can accelerate discovery, improve reproducibility, and make reasoning transparent across agents and users. More broadly, this work establishes a general framework for embedding provenance, interactivity, and explainability into the artifacts of modern research, redefining the figure not as an end product, but as an interface for discovery. For more details, please refer to the demo video available at www.llm-native-figure.com.

Problem

• LLM-native figures
• scientific discovery
• data provenance
• interactive visualization
• human-AI collaboration

Innovation

• LLM-native figures
• provenance embedding
• machine-addressable visualizations
• human-AI collaboration
• interactive scientific discovery