# DracoGPT: Extracting Visualization Design Preferences from Large Language Models

arXiv: https://arxiv.org/abs/2408.06845 · HTML: https://mlsanity.com/paper/2408.06845/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Aug 2024.

## Built on by (most cited first, 12 of 12)

- [Optimizing Sequential Multi-Step Tasks with Parallel LLM Agents](https://mlsanity.com/paper/2507.08944/) (2025)
- [ParaView-MCP: An Autonomous Visualization Agent with Direct Tool Use](https://mlsanity.com/paper/2505.07064/) (2025)
- [See or Recall: A Sanity Check for the Role of Vision in Solving Visualization Question Answer Tasks with Multimodal LLMs](https://mlsanity.com/paper/2504.09809/) (2025)
- [Exploring Multimodal Prompt for Visualization Authoring with Large Language Models](https://mlsanity.com/paper/2504.13700/) (2025)
- [A Composable Agentic System for Automated Visual Data Reporting](https://mlsanity.com/paper/2509.05721/) (2025)
- [Do Language Model Agents Align with Humans in Rating Visualizations? An Empirical Study](https://mlsanity.com/paper/2505.06702/) (2025)
- [A Generative AI System for Biomedical Data Discovery with Grammar-Based Visualizations](https://mlsanity.com/paper/2509.16454/) (2025)
- [YAC: Bridging Natural Language and Interactive Visual Exploration with Generative AI for Biomedical Data Discovery](https://mlsanity.com/paper/2509.19182/) (2025)
- [Large Language Models as Visualization Agents for Immersive Binary Reverse Engineering](https://mlsanity.com/paper/2508.13413/) (2025)
- [Toward a Machine Bertin: Why Visualization Needs Design Principles for Machine Cognition](https://mlsanity.com/paper/2602.01527/) (2026)
- [Agentic Authoring of Interactive Multiview Visualizations in Genomics](https://mlsanity.com/paper/2606.00370/) (2026)
- [MV-Bench: Benchmarking Multimodal Large Language Models for Coordinated Multi-View Interface Construction](https://mlsanity.com/paper/2607.19910/) (2026)

## Abstract (arXiv)

Trained on vast corpora, Large Language Models (LLMs) have the potential to encode visualization design knowledge and best practices. However, if they fail to do so, they might provide unreliable visualization recommendations. What visualization design preferences, then, have LLMs learned? We contribute DracoGPT, a method for extracting, modeling, and assessing visualization design preferences from LLMs. To assess varied tasks, we develop two pipelines--DracoGPT-Rank and DracoGPT-Recommend--to model LLMs prompted to either rank or recommend visual encoding specifications. We use Draco as a shared knowledge base in which to represent LLM design preferences and compare them to best practices from empirical research. We demonstrate that DracoGPT can accurately model the preferences expressed by LLMs, enabling analysis in terms of Draco design constraints. Across a suite of backing LLMs, we find that DracoGPT-Rank and DracoGPT-Recommend moderately agree with each other, but both substantially diverge from guidelines drawn from human subjects experiments. Future work can build on our approach to expand Draco's knowledge base to model a richer set of preferences and to provide a robust and cost-effective stand-in for LLMs.