Generative AI for Visualization: Opportunities and Challenges

Area of Science:

• Computer Science
• Information Visualization
• Artificial Intelligence

Background:

• Generative artificial intelligence (AI) and machine learning (ML) tools can create diverse media, prompting interest in their application to various fields.
• Speculation exists regarding AI's potential to augment or replace human activities in fields like data visualization.
• A clear understanding of generative AI's suitability for specific visualization tasks is currently lacking.

Purpose of the Study:

• To analyze the applicability of generative AI across the data visualization lifecycle.
• To identify current and emerging generative AI capabilities relevant to visualization.
• To outline the opportunities and challenges associated with integrating generative AI into visualization workflows.

Main Methods:

• Reviewing current generative AI tools and methods.
• Mapping AI capabilities to distinct phases of the visualization lifecycle (e.g., data preparation, visual encoding, interaction).
• Analyzing case studies and examples from the field.

Main Results:

• Generative AI shows potential in various visualization phases, from automated chart generation to interactive exploration.
• Current capabilities are more developed in content creation (text, images) than in complex analytical tasks.
• Challenges include ensuring accuracy, controlling output, and addressing ethical considerations.

Conclusions:

• Generative AI presents significant opportunities to augment the visualization process, particularly in automating repetitive tasks and aiding creative exploration.
• Further research is needed to address limitations and fully realize AI's potential in advanced visualization activities.
• Strategic integration of generative AI can enhance efficiency and innovation in data visualization practices.