PromptHive: Bringing Subject Matter Experts Back to the Forefront with Collaborative Prompt Engineering for Educational Content Creation
Project Overview
The document explores the innovative use of PromptHive, an open-source collaborative interface that leverages generative AI to facilitate the creation of educational content by subject matter experts. It emphasizes how PromptHive enhances the efficiency of prompt engineering, promotes collaboration among users, and ensures high content quality while alleviating the cognitive burden and time investment typically associated with content development. User studies reveal that the prompts generated through PromptHive for educational hints are of comparable quality to those created manually, indicating its effectiveness as a tool in educational settings. Overall, the findings suggest that this generative AI application not only improves the content creation process but also leads to a more positive experience for users, showcasing its potential impact on the future of education.
Key Applications
PromptHive
Context: Higher education, specifically for college-level mathematics courses
Implementation: Implemented as a collaborative interface enabling subject matter experts to create prompts for an open-source adaptive tutoring system (OATutor)
Outcomes: Reduced cognitive load by 50%, time for hint creation by a factor of 30, and generated hints comparable in quality to human-authored hints, fostering collaboration among experts.
Challenges: Initial usability issues during prompt iteration and the need for effective integration with existing workflows.
Implementation Barriers
Usability Challenges
The complexity of crafting effective prompts can be daunting for non-AI experts, leading to high metacognitive demands during prompt construction.
Proposed Solutions: Integrate scaffolding tools in the prompt authoring interface to simplify the task of prompt writing.
Integration Issues
Ensuring that generative AI tools like PromptHive integrate seamlessly into the existing workflows of subject matter experts.
Proposed Solutions: Design interfaces that allow for easy import and export of content, maintaining existing formats and structures.
Project Team
Mohi Reza
Researcher
Ioannis Anastasopoulos
Researcher
Shreya Bhandari
Researcher
Zachary A. Pardos
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Mohi Reza, Ioannis Anastasopoulos, Shreya Bhandari, Zachary A. Pardos
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