Exploring Artificial Intelligence-enhanced Electronic Design Process Logs: Empowering High School Engineering Teachers

工程设计过程（EDP）是一个通用的理论框架，经常用于教学工程，STEM，发明，甚至科学，特别是在K-12教育。虽然教育中使用的大多数edp被描述为线性或圆形，但真正的设计过程是高度创造性的，非线性的，并且经常涉及不适定的问题陈述和解决标准。这些特点使得高中工程教师尤其难以理解，因为他们往往会跳过以人为本的设计的关键要素，工程师需要花时间通过研究、访谈、先前的文献搜索、市场分析和头脑风暴来理解问题——这些步骤中思想和经验的多样性是最有价值的。此外，由于小组作业的异步性质和大班规模，很难向学生提供实时反馈，学生可能不太愿意要求对未完成的作业进行反馈。该项目将开发和试点人工智能（AI）增强的工程设计过程日志，以帮助学生导航设计过程，提供实时反馈，并鼓励对过程的每个步骤进行有意义的记录。这个项目并不打算用人工智能取代教师；相反，该项目将探索一种新的方法，即人工智能系统帮助教师创建符合EDP最佳实践的教学模块。该项目是佐治亚理工学院计算学院（GT CoC）和佐治亚理工学院综合科学、数学和计算教育中心（CEISMC）的研究人员之间的合作项目。该项目是一项以教学为重点的技术创新，代表了对人工智能增强设计教学法的早期探索。具体而言，该项目将：1)通过参与教师用户研究来改进现有的基于网络的工程设计过程日志（EDPL）； 2)设计、试点和实施一个基于人工智能的创作和辅导系统，供教师为学生和具有领域专业知识的特定项目定制反馈；3)为alpha和beta测试教师设计和提供专业发展机会；4)评估基于人工智能的EDP日志（AI-EDPL）对工程设计教学法和课堂实践的影响。AI-EDPL软件系统将使用最初为智能辅导系统开创的概念，但将其应用于定制的脚手架创建，教师制作的教学材料坚持设计过程教育学评估的最佳实践。与许多其他教育领域不同，工程设计问题的范围和解决途径各不相同，这意味着不会有一个可以向学生提供反馈的一刀切的辅导系统。本项目将研究(a)人工智能是否可以支持和支持教师创建必要的模型和知识结构，以支持和支持学习者，以及(b)教师需要什么样的专业发展才能成功开发这些模型。将采用多阶段方法，使用人机交互领域的价值敏感设计过程来开发可在教室中由教师测试的极简功能系统。为了让人工智能帮助没有太多时间摆弄软件的教师，他们必须能够用自然语言表达自己的意图，这些意图必须被自动转换成可以轻松编辑的任务模型的功能近似。该项目将以设计理论和教学法、设计文档、设计指导、设计评估和人工智能辅导的最佳实践为基础，创造一种适合高中工程设计教学的独一无二的技术。它代表了在开放式设计挑战的计算环境中提供实时反馈的首次尝试，而且它不会边缘化或削弱工程课堂上教师的作用。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The Engineering Design Process (EDP) is a general theoretical framework often used for teaching engineering, STEM, invention, and even science, particularly in K-12 education. While most EDPs used in education are depicted as linear or circular, true design processes are highly creative, non-linear, and often involve ill-posed problem statements and solution criteria. These traits make it particularly difficult for high school engineering teachers, who tend to skip over key elements of human-centered design, where an engineer takes time to understand the problem through research, interviews, prior literature searches, market analysis, and brainstorming—the steps where diversity of thought and experience are of the most value. In addition, it can be hard to provide students with real-time feedback due to the asynchronous nature of group work and large class sizes, and students may not feel comfortable asking for feedback on incomplete work. This project will develop and pilot an artificial intelligence (AI) enhanced Engineering Design Process Log to help students navigate the design process, provide real-time feedback, and encourage meaningful documentation of each step of the process. This project does not propose to replace teachers with AI; rather, the project will explore a novel approach in which AI systems assist teachers in the creation of instructional modules that adhere to EDP best practices. This project is a collaboration between researchers at Georgia Tech’s College of Computing (GT CoC) and researchers at Georgia Tech’s Center for Education Integrating Science, Mathematics and Computing (CEISMC). This project is a teaching-focused technological innovation, representing an early exploration into AI-enhanced design pedagogy. Specifically, the project will: 1) Improve upon an existing web-based Engineering Design Process Log (EDPL) by engaging in teacher user studies, 2) Design, pilot, and implement an AI-based authoring and tutoring system for teachers to customize feedback for students and for specific projects with domain expertise, 3) Design and provide professional development opportunities for alpha and beta testing teachers, and 4) Assess the impact of an AI-based EDP Log (AI-EDPL) on engineering design pedagogy and classroom practice. The AI-EDPL software system will use concepts initially pioneered for intelligent tutoring systems, but applied to scaffolding the creation of custom, teacher-made instructional materials that adhere to best practices in design process pedagogy assessment. Unlike many other educational domains, engineering design problems vary widely in scope and solution pathways, which means there will not be a one-size-fits-all tutoring system that can provide feedback to students. This project will examine (a) whether artificial intelligence can support and scaffold teachers in the creation of the necessary models and knowledge structures needed to scaffold and support learners, and, (b) what professional development teachers need to be successful in developing these models. A multi-phased approach will be used, using value-sensitive design processes from the field of human computer interaction to develop minimalist functional systems that can be tested with teachers in classrooms. In order for AI to help teachers, who do not have a lot of time to tinker with software, they must be able to express their intentions in natural language, which must be automatically converted into functional approximations of the task models that can be easily edited. This project will build on best practices in design theory and pedagogy, design documentation, design instruction, design assessment, and AI tutoring to create a one-of-a-kind technology suitable for engineering design instruction at the high school level. It represents a first attempt at providing real-time feedback in a computational setting for an open-ended design challenge, and it does so without marginalizing or diminishing the role of the instructor in the engineering classroom.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.