Skip to main content Skip to navigation

TutoAI: A Cross-domain Framework for AI-assisted Mixed-media Tutorial Creation on Physical Tasks

Project Overview

The document explores TutoAI, an innovative framework designed to enhance AI-assisted mixed-media tutorial creation, particularly for physical tasks. It identifies the complexities involved in developing mixed-media tutorials and presents TutoAI's structured approach, comprising three levels: components, models, and user interfaces. By leveraging generative AI, TutoAI streamlines the extraction and assembly of tutorial components from instructional videos, significantly improving both the user experience and the quality of the produced content. This framework not only addresses existing challenges in tutorial creation but also showcases the potential of generative AI in educational settings, making it easier for educators and learners to access and engage with high-quality, tailored instructional materials. Overall, the findings indicate that AI integration in education can lead to more effective learning tools and enhanced instructional methodologies.

Key Applications

TutoAI: A Cross-domain Framework for AI-assisted Mixed-media Tutorial Creation

Context: Creating mixed-media tutorials for physical tasks such as cooking, hardware assembly, etc.

Implementation: TutoAI identifies common tutorial components, assembles AI models for component extraction, and provides a user-friendly interface for creators to review and edit AI-generated components.

Outcomes: Higher quality tutorial components compared to baseline models, improved user experience in tutorial creation, and successful integration into creator workflows.

Challenges: Existing models are often domain-specific, and integrating AI with mixed-media tutorial creation involves complexities such as managing multi-modal data.

Implementation Barriers

Technical Barrier

Integrating AI into mixed-media tutorial creation is complex due to the variety of models and multi-modal data involved.

Proposed Solutions: Developing general guidelines for model assembly and ensuring a systematic approach to combining AI techniques.

User Experience Barrier

Users may feel overwhelmed by the information presented and struggle with managing multiple modalities in tutorial creation.

Proposed Solutions: Implementing UI design principles that simplify user interactions and allow for sequential task management.

Project Team

Yuexi Chen

Researcher

Vlad I. Morariu

Researcher

Anh Truong

Researcher

Zhicheng Liu

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Yuexi Chen, Vlad I. Morariu, Anh Truong, Zhicheng Liu

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies