Explanation as Question Answering based on a Task Model of the Agent's Design
Project Overview
The document outlines the implementation of Skillsync, an innovative AI platform aimed at enhancing worker upskilling and reskilling by connecting educational institutions with businesses. A key feature, AskJill, serves as a question-answering tool that clarifies Skillsync's functionalities and terminology, thereby improving user engagement and understanding. The development process employs a human-centered design strategy, utilizing participatory design techniques with focus groups to pinpoint user inquiries and requirements effectively. The framework of the Task-Method-Knowledge (TMK) model is introduced to organize the acquired knowledge and to bolster the transparency and explainability of AI systems within educational settings. Overall, the initiative highlights the pivotal role of generative AI in education, showcasing its potential to facilitate learning and skill development, while ensuring that the systems remain user-friendly and aligned with the needs of learners and educators alike. The findings emphasize the importance of integrating user feedback in the design of AI tools to create more effective educational resources that can adapt to the evolving landscape of workforce requirements.
Key Applications
Skillsync and AskJill
Context: Skillsync is used in the context of workforce development, targeting companies and educational institutions seeking to align training with industry needs.
Implementation: Skillsync employs a TMK model to structure the design and functionality of AskJill, which answers user questions regarding training requests and educational proposals.
Outcomes: Improved user understanding of the Skillsync platform, enhanced transparency of AI recommendations, and increased efficiency in matching jobs with educational programs.
Challenges: Ensuring the accuracy of responses, addressing user questions that fall outside AskJill's current knowledge base, and maintaining user trust in AI recommendations.
Implementation Barriers
Technical barrier
Complexity of AI design makes it challenging to generate comprehensive explanations.
Proposed Solutions: Utilizing a TMK model to clarify tasks and knowledge, and conducting participatory design to gather user input on necessary explanations.
User acceptance barrier
Users may be skeptical about the trustworthiness and transparency of AI systems.
Proposed Solutions: Building trust through clear explanations of AI processes and user feedback mechanisms to improve the system continuously.
Project Team
Ashok Goel
Researcher
Harshvardhan Sikka
Researcher
Vrinda Nandan
Researcher
Jeonghyun Lee
Researcher
Matt Lisle
Researcher
Spencer Rugaber
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Ashok Goel, Harshvardhan Sikka, Vrinda Nandan, Jeonghyun Lee, Matt Lisle, Spencer Rugaber
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