Learning Mathematical Concepts and Computational Thinking through Explainable Artificial Intelligence in a Simulation-based Learning Environment

由于STEM学科中计算的强大创新和应用，STEM+C计划支持跨学科和跨学科方法的研究和开发，以将计算整合到正式和非正式环境中的PreK-12学生的STEM教学和学习中。该项目将促进数学教育在人工智能（AI）和高中生计算思维技能背景下的整合。AI已经在我们的日常生活中无处不在。从虚拟助手到自动驾驶汽车，从诊断疾病到建造房屋，今天的许多学生将继续在涉及人工智能或受人工智能影响的领域工作。精通人工智能语言是一支能够继续创新并支持人工智能技术基础设施和生态系统的劳动力队伍的关键。AI建立在数学的基础上。通过说明如何在强大的人工智能工具中使用数学概念来解决计算问题，通过人工智能和计算学习数学可以成为一种激励和教育工具，以说明从K-12 STEM教育到中学后STEM教育，再到STEM职业的途径。今天的大多数人工智能决策过程对于非人工智能专家，甚至对于一些人工智能专家来说都是一个“黑匣子”。可解释人工智能（一种新兴的智能人机界面，使人工智能算法的决策对用户透明）的最新进展，为高中生提供了一个机会，使人工智能机器学习概念变得容易理解。该项目在基于模拟的学习环境中采用可解释的人工智能，学生可以遵循指导的人机团队探索，学习如何创建问题的抽象，利用人工智能算法自动生成解决方案的过程，分析结果，然后提高他们的解决方案的性能。 研究人员将反复引入挑战问题和场景，鼓励学生修改机器人的决策，学生诊断，修改，测试和分析，并为机器人创建新的功能。拟议中的项目与来自弗吉尼亚州和加州的三所高中合作。综合学习内容将与高中教师一起开发，重点是高中数学，人工智能和计算思维技能之间的联系。该项目旨在回答以下问题：在高中学习者的使用-创建-修改框架中，交互式可解释AI设计选择和解释在多大程度上有助于理解数学和AI。研究将解决（1）评估使用-创建-修改方法及其对计算思维和使用人工智能的自我效能的影响，（2）评估对数学和人工智能的理解，以及解释的不同用途，以促进学习，以及（3）享受和参与人机模拟，以及在这个人类的行为中，技术交互式学习环境能够框高中水平的数学学习及其与计算思维的整合。该奖项反映了NSF的法定使命，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，认为值得支持。

As a result of the powerful innovation and application of computing in STEM disciplines, the STEM+C program supports research and development of interdisciplinary and transdisciplinary approaches to the integration of computing within STEM teaching and learning for preK-12 students in both formal and informal settings. This project will advance the integration of the education of math within the context of artificial intelligence (AI) and computational thinking skills for high school students. AI has become ubiquitous in our everyday lives. From virtual assistants, to self-driving cars, from diagnosing disease, to building houses, many of today's students will go on to work in fields that involve or are influenced by AI. Being proficient in the language of AI is key to a workforce that will be able to continue to innovate and to support the AI-powered technology infrastructure and eco-system. AI builds on the foundation of mathematics. By illustrating how math concepts can be used in powerful AI tools to solve computational problems, learning math through AI and computation can be a motivational and educational vehicle to illustrate the pathway from K-12 STEM education, to post-secondary STEM education, and later to STEM careers. Most of the AI decision-making process today is a "black box" to non-AI experts, and even to some AI experts. Recent advances in explainable AI, an emerging intelligent human-computer interface that makes the decision-making of AI algorithms transparent to users, creates an opportunity to make AI machine learning concepts accessible to high school students. This proposed project employs explainable AI within a simulation-based learning environment where students follow guided human-robot team explorations to learn how to create abstractions of a problem, utilize AI algorithms to automate the process of solution generation, analyze the outcome, and then improve the performance of their solution. Researchers will iteratively introduce challenge problems and scenarios to encourage students to modify the robot's decision-making where students diagnose, revise, test and analyze, and create new capabilities for the robot. The proposed project partners with three high schools from Virginia and California. The integrated learning content will be developed with high school teachers, focusing on the connections between high school math, AI and computational thinking skills. The proposed project aims to answer the question of to what extent interactive explainable AI design choices and explanations contribute to understanding of math and AI in a use-create-modify framework for high school learners. Research studies will address (1) assessment of the use-create-modify approach and its impact on self-efficacy for computational thinking and use of AI, (2) assessment of understanding of math and AI, and the different uses of explanations to promote learning, and (3) enjoyment and engagement with the human-robot simulation, and to what extent enactment in this human-technology interactive learning environment is able to frame high school level learning of math and its integration with computational thinking.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.