[Paper Review] Where Are We So Far? Understanding Data Storytelling Tools from the Perspective of Human-AI Collaboration

Haotian Li, Yun Wang+1 more|arXiv (Cornell University)|Sep 27, 2023

Big Data and Business Intelligence4 citations

TL;DR

This paper presents a systematic framework to analyze human-AI collaboration in data storytelling tools, categorizing their functionality across four workflow stages—analysis, planning, implementation, and communication—and mapping roles such as creator, assistant, optimizer, and reviewer. The study reveals that human-AI collaboration has gained momentum since 2019, with recurring patterns in role distribution and workflow integration, offering design insights and identifying future research opportunities in automation, communication, and multi-agent collaboration.

ABSTRACT

Data storytelling is powerful for communicating data insights, but it requires diverse skills and considerable effort from human creators. Recent research has widely explored the potential for artificial intelligence (AI) to support and augment humans in data storytelling. However, there lacks a systematic review to understand data storytelling tools from the perspective of human-AI collaboration, which hinders researchers from reflecting on the existing collaborative tool designs that promote humans' and AI's advantages and mitigate their shortcomings. This paper investigated existing tools with a framework from two perspectives: the stages in the storytelling workflow where a tool serves, including analysis, planning, implementation, and communication, and the roles of humans and AI in each stage, such as creators, assistants, optimizers, and reviewers. Through our analysis, we recognize the common collaboration patterns in existing tools, summarize lessons learned from these patterns, and further illustrate research opportunities for human-AI collaboration in data storytelling.

Motivation & Objective

• To understand how human-AI collaboration is structured in existing data storytelling tools across different stages of the storytelling workflow.
• To identify recurring collaboration patterns between humans and AI by analyzing roles such as creator, assistant, optimizer, and reviewer in each stage.
• To provide a systematic review of human-AI collaborative data storytelling tools from 2010 to 2023, highlighting research trends and design gaps.
• To inform future tool design by summarizing lessons learned and identifying underexplored research opportunities in automation, communication, and multi-agent collaboration.
• To establish a sustainable research resource through an online tool browser for tracking the evolution of human-AI collaborative data storytelling systems.

Proposed method

• The authors constructed a paper corpus from VIS and HCI literature (2010–2023) focusing on human-AI collaborative data storytelling tools.
• They applied a dual-dimensional framework: (1) the stages of the storytelling workflow—analysis, planning, implementation, and communication—and (2) the roles of human and AI collaborators—creator, assistant, optimizer, reviewer.
• Each tool was coded based on its coverage of stages and the defined roles played by humans and AI in each stage.
• The analysis identified common collaboration patterns, design trends, and disparities across stages and roles.
• The study used qualitative synthesis to extract design principles, challenges, and research opportunities from the coded data.
• The authors developed and maintain an online tool browser at https://human-ai-universe.github.io/data-storytelling to track and update the evolving landscape of these tools.

Experimental results

Research questions

• RQ1 How do human-AI collaborative data storytelling tools distribute roles across the stages of data storytelling—analysis, planning, implementation, and communication?
• RQ2 What are the dominant collaboration patterns between humans and AI in existing data storytelling tools, and how do they vary across different stages?
• RQ3 What design principles and challenges emerge from analyzing human-AI collaboration in data storytelling tools?
• RQ4 How has the research focus on human-AI collaboration in data storytelling evolved from 2010 to 2023?
• RQ5 What future research opportunities exist for enhancing automation, communication, and multi-agent collaboration in human-AI data storytelling systems?

Key findings

• Human-AI collaborative data storytelling tools have gained significant research interest since 2019, following a period of dominance by purely manual tools from 2016 to 2018.
• The most frequent collaboration pattern involves AI acting as an assistant in data analysis and visualization generation, while humans serve as creators and reviewers in planning and communication stages.
• AI is most commonly used to support data fact discovery, visualization recommendation, and feedback generation, while humans contribute domain knowledge, creativity, and contextual awareness.
• The roles of optimizer and reviewer are less frequently implemented, indicating underutilization of AI for quality assessment and refinement in storytelling workflows.
• There is a notable gap in tools supporting multi-human and multi-AI collaboration, suggesting an emerging need for coordination mechanisms in complex storytelling environments.
• The study identifies communication methods between humans and AI—such as programming, sketching, and gesture—as under-investigated areas with high potential for future research.