Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Advancing GUI for Generative AI: Charting the Design Space of Human-AI Interactions through Task Creativity and Complexity

Zijian Ding

TL;DR

Advancing GUI for Generative AI examines how to design user interfaces that couple human expertise with Generative AI across creative and analytical tasks. It proposes a three-part plan—fixed-scope news headline curation, atomic cross-domain analogy generation, and complex data visualization—mapped to a human-AI interaction taxonomy to guide GUI design. The studies suggest guidance+selection often yields the best balance of quality and efficiency, while human oversight remains essential for safety, bias, and accuracy; complex tasks benefit from interactive, design-like canvases enabling rapid prototyping and multi-modal reasoning. The work provides a concrete design roadmap for mixed-initiative, multi-modal GUIs that enhance interpretability, collaboration, and control in Generative AI-enabled workflows.

Abstract

Technological progress has persistently shaped the dynamics of human-machine interactions in task execution. In response to the advancements in Generative AI, this paper outlines a detailed study plan that investigates various human-AI interaction modalities across a range of tasks, characterized by differing levels of creativity and complexity. This exploration aims to inform and contribute to the development of Graphical User Interfaces (GUIs) that effectively integrate with and enhance the capabilities of Generative AI systems. The study comprises three parts: exploring fixed-scope tasks through news headline generation, delving into atomic creative tasks with analogy generation, and investigating complex tasks via data visualization. Future work aims to extend this exploration to linearize complex data analysis results into narratives understandable to a broader audience, thereby enhancing the interpretability of AI-generated content.

Advancing GUI for Generative AI: Charting the Design Space of Human-AI Interactions through Task Creativity and Complexity

TL;DR

Advancing GUI for Generative AI examines how to design user interfaces that couple human expertise with Generative AI across creative and analytical tasks. It proposes a three-part plan—fixed-scope news headline curation, atomic cross-domain analogy generation, and complex data visualization—mapped to a human-AI interaction taxonomy to guide GUI design. The studies suggest guidance+selection often yields the best balance of quality and efficiency, while human oversight remains essential for safety, bias, and accuracy; complex tasks benefit from interactive, design-like canvases enabling rapid prototyping and multi-modal reasoning. The work provides a concrete design roadmap for mixed-initiative, multi-modal GUIs that enhance interpretability, collaboration, and control in Generative AI-enabled workflows.

Abstract

Technological progress has persistently shaped the dynamics of human-machine interactions in task execution. In response to the advancements in Generative AI, this paper outlines a detailed study plan that investigates various human-AI interaction modalities across a range of tasks, characterized by differing levels of creativity and complexity. This exploration aims to inform and contribute to the development of Graphical User Interfaces (GUIs) that effectively integrate with and enhance the capabilities of Generative AI systems. The study comprises three parts: exploring fixed-scope tasks through news headline generation, delving into atomic creative tasks with analogy generation, and investigating complex tasks via data visualization. Future work aims to extend this exploration to linearize complex data analysis results into narratives understandable to a broader audience, thereby enhancing the interpretability of AI-generated content.
Paper Structure (10 sections, 3 figures)

This paper contains 10 sections, 3 figures.

Table of Contents

  1. Introduction
  2. Related work
  3. Fixed-scope Content Curation Tasks
  4. Atomic Creative Tasks
  5. Complex and Interdependent Tasks
  6. Preliminary Studies and Results
  7. Study 1 on Fixed-Scope Content Curation Tasks: News Headline Generation
  8. Study 2 on Atomic Creative Tasks: Cross-Domain Analogy Generation
  9. Study 3 on Complex and Interdependent Tasks: Data Visualization
  10. Implications and future work

Figures (3)

  • Figure 1: Study 1 on fixed-scope content curation tasks - interface for human-AI news headline co-creation for guidance + selection + post-editing condition: (A) news reading panel, (B) perspectives (keywords) selection panel (multiple keywords can be selected), (C) headline selection panel with post-editing capability, and (D) difficulty rating slider.
  • Figure 2: Study 2 on atomic creative tasks - our learning prompts based on Duncker and Lees' radiation problem dunckerProblemsolving1945 for generating cross-domain analogies.
  • Figure 3: Study 3 on complex and interdependent tasks - data analysis results in design-like canvas environment: results (left) and study setting (right).