Artificial Intelligence in STEM Education: The Paradigmatic Shifts in Research, Education, and Technology

Description

Artificial intelligence (AI) opens new opportunities for STEM education in K-12, higher education, and professional education contexts. This book summarizes AI in education (AIED) with a particular focus on the research, practice, and technological paradigmatic shifts of AIED in recent years. The 23 chapters in this edited collection track the paradigmatic shifts of AIED in STEM education, discussing how and why the paradigms have shifted, explaining how and in what ways AI techniques have ensured the shifts, and envisioning what directions next-generation AIED is heading in the new era. As a whole, the book illuminates the main paradigms of AI in STEM education, summarizes the AI-enhanced techniques and applications used to enable the paradigms, and discusses AI-enhanced teaching, learning, and design in STEM education. It provides an adapted educational policy so that practitioners can better facilitate the application of AI in STEM education. This book is a must-read for researchers, educators, students, designers, and engineers who are interested in the opportunities and challenges of AI in STEM education. Cover Half Title Series Page Title Page Copyright Page Table of Contents List of Contributors Editor Biographies Section I AI-Enhanced Adaptive, Personalized Learning Chapter 1 Artificial Intelligence in STEM Education: Current Developments and Future Considerations 1.1 Introduction 1.2 Paradigmatic Shifts of AI in STEM Education 1.2.1 Paradigm One: AI-Directed STEM Education 1.2.2 Paradigm Two: AI-Supported STEM Education 1.2.3 Paradigm Three: AI-Empowered STEM Education 1.3 Discussion and Future Considerations 1.4 Structure of the Book 1.5 Conclusions Acknowledgment References Chapter 2 Towards a Deeper Understanding of K-12 Students’ CT and Engineering Design Processes 2.1 Introduction 2.2 Background and Motivation 2.2.1 21st-Century K-12 Classrooms 2.2.2 Understanding Students’ STEM Learning Strategies 2.2.2.1 Computational Modeling Strategies 2.2.2.2 Engineering Design Strategies 2.3 SPICE Curriculum and Learning Environment 2.3.1 System Design Perspectives 2.3.1.1 Evidence-Centered Design 2.3.1.2 Coherence across Modeling Representations 2.3.1.3 Domain-Specific Modeling Languages 2.3.2 SPICE Learning Trajectory and Curriculum 2.4 Methods 2.4.1 Implementation 2.4.2 Data Sources 2.4.3 Analysis Methods 2.5 Results 2.5.1 The Role of Computational Thinking 2.5.1.1 Effects on Computational Modeling 2.5.1.2 Effects on Engineering Design 2.5.1.3 Effects on Posttest Scores 2.5.2 Learning with Multiple Representations 2.5.3 Impact of Strategy Use on Learning 2.5.3.1 Computational Modeling Strategies 2.5.3.2 Engineering Design Strategies 2.6 Discussion and Future Directions Notes Bibliography Chapter 3 Intelligent Science Stations Bring AI Tutoring into the Physical World 3.1 Introduction 3.2 Intelligent Science Stations: A Guided AI Mixed-Reality Interactive Experience 3.2.1 Physical Setup and AI Computer Vision Technology 3.2.2 Scenario 3.3 AI in the 3D Physical World versus on Flat Screen 3.4 AI Guidance Matters: Making Science Exhibits Yield Better Learning and More Engagement 3.5 The Kind of AI Guidance Matters: Scaffolded Science Inquiry Produces Better Engineering 3.6 Discussion and Next Steps Note References Chapter 4 Adaptive Support for Representational Competencies during Technology-Based Problem-Solving in STEM 4.1 Introduction 4.2 Literature Review 4.2.1 Sense-Making Competencies 4.2.2 Perceptual Fluency 4.3 Empirical Studies 4.3.1 Combining Sense-Making Support and Perceptual-Fluency Support 4.3.2 How Should Sense-Making and Perceptual-Induction Activities Be Sequenced? 4.3.3 Should Representational-Competency Supports Adapt to Students’ Current Knowledge Level? 4.4 Discussion Note References Chapter 5 Teaching STEM Subjects in Non-STEM Degrees: An Adaptive Learning Model for Teaching Statistics 5.1 Introduction 5.2 State of the Art 5.2.1 A Hierarchical Structure for Statistical Knowledge 5.2.2 Knowledge Space Theory 5.2.3 Dublin Descriptors 5.3 Methodology 5.3.1 Multidimensional IRT Model 5.3.2 Archetypal Analysis 5.4 Implementation 5.4.1 App Interface and Workflow 5.4.2 User Model and Data Processing 5.5 Simulation Data Results 5.5.1 Design of the Study 5.5.2 Results 5.5.3 Preliminary Application on Real-world Data 5.6 Discussion 5.7 Limitations and Future Work 5.8 Conclusion References Chapter 6 Removing Barriers in Self-Paced Online Learning through Designing Intelligent Learning Dashboards 6.1 Introduction 6.1.1 Intelligent Learning Dashboards 6.1.2 Automatic Generation of Formative Assessments 6.1.3 Automatic Analysis of Students’ Engagement 6.2 Overview of the Intelligent Learning Dashboard Framework 6.3 Adaptive Formative Assessments 6.4 Automatic Detection of Students’ Engagement 6.5 Learning Analytic Dashboards 6.6 Conclusions and Future Work Acknowledgment References Section II AI-Enhanced Adaptive Learning Resources Chapter 7 PASTEL: Evidence-based Learning Engineering Methods to Facilitate Creation of Adaptive Online Courseware 7.1 Introduction 7.2 CyberBook 7.3 PASTEL 7.3.1 WATSON 7.3.2 SMART 7.3.3 QUADL 7.3.4 RAFINE 7.3.5 RADARS 7.4 Evaluation 7.4.1 SMART 7.4.2 QUADL 7.4.3 RAFINE 7.4.4 RADARS 7.5 Conclusion Notes References Chapter 8 A Technology-Enhanced Approach for Locating Timely and Relevant News Articles for Context-Based Science Education 8.1 Literature Review 8.1.1 Science News Articles and Assessment in Context-Based Science Education 8.1.2 Automated Curriculum Alignment of Learning Resources in Science Education 8.1.3 Topic Modeling and Latent Dirichlet Allocation 8.1.4 Present Study 8.2 Methods 8.2.1 Data 8.2.1.1 Stage 1: Science Article Preprocessing and Vectorization 8.2.1.2 Stage 2: Science Article Topic Modeling 8.2.1.3 Stage 3: Topic Structure Prediction and Evaluation of the Units of Study 8.2.1.4 Stage 4: Science Article Recommendation Based on Topic Similarity 8.3 Results 8.3.1 Science Article Topic Analysis 8.3.2 Topic Prediction and Topic Evaluation Units of Study 8.3.3 Science Article Recommendation and Item Generation 8.4 Conclusions and Discussion Appendix A Appendix B Appendix C Appendix D Notes References Chapter 9 Adaptive Learning Profiles in the Education Domain 9.1 Introduction 9.2 Academic Analytics 9.2.1 Decision Support in Education: Loyalty and Dropout 9.2.2 Academic Analytics at Politecnico di Torino 9.2.3 Research Questions 9.2.4 The Information System Course 9.2.4.1 Framework 9.2.4.2 Classroom 9.2.4.3 Course Delivery Organization 9.2.4.4 Assessment 9.2.4.5 Course Management: Student Behavior and Assessment 9.3 Academic Analytics Platform 9.4 Information Systems Course Data 9.4.1 Dataset 9.4.2 Dataset Initial Exploration 9.4.3 Clustering Algorithm Overview and Benchmark 9.4.3.1 DBSCAN 9.4.3.2 k-Means and k-Medoids 9.4.4 Performance-Based Clustering Visualization 9.4.5 Association Rules 9.4.5.1 Cluster 0: Exam Passed – ‘Advanced’ Python/UML Section, ‘Intermediate’ MC Section, and ‘Intermediate’ Project Work 9.4.5.2 Cluster 1: Exam Passed – ‘Advanced’ Python/UML Section, ‘Expert’ MC Section, and ‘Intermediate’ Project Work 9.4.5.3 Cluster 2: Exam Failed –‘Basic’ MC Section, ‘Being Developed’ Python/UML Section, and No Project Work 9.4.5.4 Cluster 3: Exam Passed/Failed –‘Intermediate’ Python/UML Section, ‘Advanced’ MC Section, and ‘Intermediate’ Project Work 9.4.5.5 Cluster 4: Exam Failed – ‘Being Developed’ Python/UML and MC Sections and ‘Intermediate’ Project Work 9.4.5.6 Cluster 5: Exam Failed – ‘Being Developed’ Python/UML and MC Sections and No Project Work 9.4.5.7 Cluster 6: Exam Passed/Failed – All ‘Intermediate’ Sections 9.4.5.8 Cluster 7: Exam Passed – ‘Expert’ Python/UML and MC Sections and 'Intermediate’/‘Advanced’ Project Work 9.4.5.9 Cluster 8: Exam Failed – ‘Being Developed’ Python/UML Section, ‘Basic’/ ‘Intermediate’ MC Section, and ‘Intermediate’ Project Work 9.4.6 Discussion on the Results 9.4.7 Research Questions and Answers 9.5 Conclusions and Future Work References Section III AI-Supported Instructor Systems and Assessments for AI and STEM Education Chapter 10 Teacher Orchestration Systems Supported by AI: Theoretical Possibilities and Practical Considerations 10.1 Classroom Orchestration 10.2 Artificial Intelligence for Classroom Orchestration 10.3 Research Context 10.4 The Role of AI in Orchestration Assistant Design 10.5 Classroom Context Interviews 10.6 Orchestration Assistant Design 10.6.1 During Class versus Beyond Class Assistance 10.6.2 Automated Systems versus Teacher Agency 10.6.3 Detailed Information versus Actionable Information 10.6.4 Classroom Management versus Ambitious Learning Practices 10.7 Discussion and Future Research References Chapter 11 The Role of AI to Support Teacher Learning and Practice: A Review and Future Directions 11.1 Background 11.1.1 Teacher Learning Perspectives 11.1.2 Technologies for Teacher Learning 11.1.2.1 AI-Based Teacher Tools 11.1.3 Design Guidance for AIEd Tools 11.1.4 The ICAP Framework 11.2 ICAP and AI-Based Technologies for Teacher Learning 11.2.1 Passive AIEd Technologies for Teacher Learning 11.2.2 Active AIEd Technologies for Teacher Learning 11.2.3 Constructive AIEd Technologies for Teacher Learning 11.2.4 Interactive AIEd Technologies for Teacher Learning 11.3 Discussion 11.4 Limitations 11.5 Conclusion Acknowledgment References Chapter 12 Learning Outcome Modeling in Computer-Based Assessments for Learning 12.1 Psychometric Measurement for Learning Outcome Modeling 12.1.1 Classical Test Theory 12.1.2 Item Response Theory 12.1.3 Cognitive Diagnosis 12.2 Bayesian Networks 12.3 Bayesian Knowledge Tracing 12.4 Additive Factors Model 12.5 Deep Learning for Learning Outcome Modeling 12.5.1 Deep Knowledge Tracing 12.5.2 Other Deep Learning Approaches for Learning Outcome Modeling 12.6 Collaborative Filtering for Learning Outcome Modeling 12.6.1 Matrix Factorization 12.6.2 Collaborative Filtering-Based Approaches for Learning Outcome Modeling 12.6.3 Deep Learning–Based Collaborative Filtering 12.7 An Overview of Approaches for Learning Outcome Modeling 12.8 Conclusion References Chapter 13 Designing Automated Writing Evaluation Systems for Ambitious Instruction and Classroom Integration 13.1 Introduction 13.1.1 Overview of eRevise System 13.1.2 Chapter Overview 13.2 Considerations for Automated Writing Evaluation (AWE) System Design 13.2.1 Authentic Tasks That Aid Ambitious Teaching 13.2.2 Features of a Construct Are Identified and Named 13.2.3 Information Is at a ‘Grain Size’ to Support Learning 13.2.4 Underlying Algorithms Are Fair 13.2.5 System Supports Instructional Interactions and Discourse 13.3 Designing to Facilitate Widespread Adoption of AWE Systems and Integration in Classroom Routines 13.3.1 Education Policies 13.3.2 Values and Goals Held by School Leaders 13.3.3 Values and Goals Held by Teachers 13.4 Concluding Thoughts References Section IV Learning Analytics and Educational Data Mining in AI and STEM Education Chapter 14 Promoting STEM Education through the Use of Learning Analytics: A Paradigm Shift 14.1 Introduction 14.2 Learning Analytics as an Enabler for a Paradigm Shift in STEM Education 14.2.1 The Advancement of Data Collection Techniques 14.2.2 The Expansion of Data Analysis Methods 14.2.3 The Emergence of Innovative Visualization Strategies 14.3 Facilitating Change in STEM Education with Theory-Driven Learning Analytics 14.4 Case Example: Examining Students’ STEM Learning with SRL Theory 14.4.1 Theory-Driven Data Collection 14.4.2 Theory-Driven Data Analysis and Visualization 14.4.3 Theory-Driven Interpretation of Results 14.5 The Future of STEM Education with Learning Analytics References Chapter 15 Using Learning Analytics to Understand Students’ Discourse and Behaviors in STEM Education 15.1 Introduction 15.2 STEM Education 15.3 Technology-Enhanced Environments to Support STEM Education 15.3.1 Teaching Teamwork for Collaborative Problem-Solving 15.3.2 Energy3D for Engineering Design 15.4 Learning Analytics to Analyze Generated Data 15.4.1 Analytical Methods 15.5 Case Studies 15.5.1 Science 15.5.1.1 Transformative and Non-Transformative Discourse 15.5.1.2 Self-Regulation 15.5.1.3 Multi-Faceted Engagement 15.5.2 Engineering 15.5.3 Mathematics 15.6 Research Gaps and Future Trends 15.7 Conclusion References Chapter 16 Understanding the Role of AI and Learning Analytics Techniques in Addressing Task Difficulties in STEM Education 16.1 Introduction 16.2 The Multifaceted Aspects of Task Difficulty 16.3 LA and AI in Digital Learning Environments 16.3.1 Virtual Labs 16.3.2 Serious Games and Simulations 16.4 Opportunities 16.4.1 Real-Time Feedback and Intervention 16.4.2 Personalized Adaptive Learning 16.4.3 Leveraging the Advances in Sensor Technology 16.5 Challenges 16.5.1 Task Difficulty and Its Operationalization 16.5.2 Data Ethics and Student Privacy 16.5.3 What Data to Collect and How 16.5.4 Confounding Factors 16.6 Conclusion Note References Chapter 17 Learning Analytics in a Web3D Based Inquiry Learning Environment 17.1 Introduction 17.1.1 Learning Analytics 17.1.2 Web3D Technology 17.1.2.1 What Is Web3D 17.1.2.2 Implementation Technologies of Web3D 17.1.3 Web3D-Based Inquiry Learning Environment 17.1.4 Web3D and Learning Analytics 17.2 Theoretical Basis of Learning Analytics 17.2.1 Knowledge Graph 17.2.1.1 The Origin of the Knowledge Graph 17.2.1.2 Educational Knowledge Graph 17.2.1.3 Theoretical Basis of Educational Knowledge Graph 17.2.1.4 Principles of Educational Knowledge Graph Construction 17.2.2 Learner Profile 17.2.2.1 What Is the Learner Profile 17.2.2.2 Research Status of Learner Profile 17.2.3 Theory and Record of Learning Behavior 17.2.3.1 Theoretical Basis 17.2.3.2 Standards for Recording Behavior Data 17.3 Technical Path of Learning Analytics 17.3.1 Realization Methods of Knowledge Graph 17.3.1.1 Construction of Educational Knowledge Graph 17.3.1.2 Storage and Presentation of Educational Knowledge Graph 17.3.2 How to Create a Learner Profile 17.3.2.1 Knowledge Level Assessment 17.3.2.2 Learning Behavior Assessment 17.3.2.3 Emotion and Attitude Assessment 17.3.3 Definition and Analytics Method of Learning Behavior 17.3.3.1 Definition of Learning Behavior Analytics 17.3.3.2 Analytics Method of Learning Behavior 17.4 Concrete Realization of Learning Analysis in Web3D Based Inquiry Learning Environment 17.4.1 Construction of Knowledge Graph of Chemistry Experiments in Middle School 17.4.2 Generation and Presentation of Learner Profile in Web-Based Inquiry Learning Environment 17.4.3 Analysis and Evaluation of Inquiry Learning Behavior 17.5 Summary and Outlook References Chapter 18 On Machine Learning Methods for Propensity Score Matching and Weighting in Educational Data Mining Applications 18.1 Introduction 18.2 Methods 18.2.1 Propensity Score 18.2.1.1 Assumptions for Propensity Score-Based Methods 18.2.1.2 Propensity Score Matching 18.2.1.3 Inverse Probability of Treatment Weighting 18.2.2 Random Forest 18.2.3 Ensemble Learning 18.3 Simulation Study 18.3.1 Data Generation 18.3.1.1 Generating Covariates (X) 18.3.1.2 Generating Treatment Assignment Indicator (Z) 18.3.1.3 Generating Outcome (Y) 18.3.2 Simulation Study Results 18.4 Students’ Success Case Study 18.5 Discussion Acknowledgment References Chapter 19 Situating AI (and Big Data) in the Learning Sciences: Moving toward Large-Scale Learning Sciences 19.1 Introduction 19.2 A Landscape of Learning Sciences and Some Challenges It Faces 19.2.1 Outcomes 19.2.2 Student Factors 19.2.3 Contextual Factors 19.2.4 Replication Crisis? (Or Maybe Context Matters) 19.3 AI and Its Affordances for the Learning Sciences 19.3.1 Deep Student Models 19.3.2 Causal Learning Outcome Models 19.3.3 Natural Language Processing 19.3.4 Sensor-Free Student Factor Measures 19.3.5 Instructional Policy Learning 19.4 Promoting Equity 19.5 Conclusion Acknowledgments References Chapter 20 Linking Natural Language Use and Science Performance 20.1 Introduction 20.1.1 Language Ability and Science Scores 20.1.2 Computer-Based Science Education 20.1.3 The Current Study 20.2 Method 20.2.1 Procedure 20.2.2 Participants 20.2.3 Data Collected 20.2.3.1 Pretest/Posttest Assessments 20.2.3.2 Individual Differences Data 20.2.3.3 Verbal Data 20.2.4 Transcriptions 20.2.5 Linguistic Variables 20.2.5.1 TAALES 20.2.5.2 TAACO 20.2.5.3 SEANCE 20.2.6 Statistical Analysis 20.3 Results 20.3.1 Non-Linguistic Model 20.3.2 Linguistic Model 20.3.3 Full Model 20.4 Discussion 20.5 Conclusion Acknowledgments References Section V Other Topics in AI and STEM Education Chapter 21 Quick Red Fox: An App Supporting a New Paradigm in Qualitative Research on AIED for STEM 21.1 Introduction 21.2 Design 21.2.1 Interview Triggers 21.2.2 Server-Side Platform 21.2.3 Client Side 21.2.3.1 Set-Up and Login 21.2.3.2 Presentation of Student and Trigger Information 21.2.3.3 Interview Recordings and Notes 21.2.3.4 Moving On (Next, Skip, End) 21.2.3.5 Data 21.3 Case Study 21.3.1 Betty’s Brain 21.3.2 Developing Interview Triggers 21.3.3 Procedure 21.3.4 Data 21.3.5 Data Coding 21.3.6 Impact on Scholarly Work 21.4 General Discussion and Conclusions 21.4.1 Summary 21.4.2 Applications 21.4.3 Limitations 21.4.4 Future Development Acknowledgments Note References Chapter 22 A Systematic Review of AI Applications in Computer-Supported Collaborative Learning in STEM Education 22.1 Introduction and Review of Previous Work 22.2 Methodology 22.2.1 Research Purpose and Questions 22.2.2 Searching the Literature 22.2.3 Selecting Studies 22.2.4 Data Extraction 22.2.5 Data Synthesis and Analysis 22.3 Results 22.3.1 RQ1: The Overall Trend 22.3.1.1 Type of Research 22.3.1.2 Educational Contexts 22.3.1.3 Research Focus 22.3.1.4 Research Methods 22.3.2 RQ2: AI Techniques Used to Support Group Formation 22.3.2.1 Group Size and Grouping Mechanisms 22.3.2.2 AI Techniques and Evaluation 22.3.3 RQ3: AI-Enabled Systems to Support Student Interactions 22.3.3.1 Student Interaction Design 22.3.3.2 Student Interaction Evaluation 22.4 Discussions and Implications 22.4.1 Overall Trend of AI Applications in CSCL Research 22.4.1.1 Type of Research Paper 22.4.1.2 Research Focus 22.4.1.3 Educational Contexts 22.4.1.4 Research Design 22.4.2 AI Applications to Support Group Formation 22.4.2.1 Issues of Design 22.4.2.2 Issues of Evaluation 22.4.3 AI Applications to Support Student Interactions 22.4.3.1 Issues of Design 22.4.3.2 Issues of Evaluation 22.5 Conclusions References Chapter 23 Inclusion and Equity as a Paradigm Shift for Artificial Intelligence in Education 23.1 Inclusion and Equity as a Paradigm Shift for Artificial Intelligence in Education 23.2 AI and DEI: A Bidirectional Relationship 23.2.1 AI for DEI: How Can the Principles and Methods of Artificial Intelligence Support Diversity, Equity, and Inclusion? 23.2.2 Inclusivity in STEM Introductory Courses 23.2.2.1 Performance in Introductory STEM Courses 23.2.2.2 Active Learning in Introductory STEM Courses 23.2.3 Collaboration and Discourse 23.2.4 Learning Assessments 23.2.5 DEI for AI: How Can the Principles of Diversity, Equity, and Inclusion Transform Artificial Intelligence in Education? 23.2.6 Person-Centered Variables, Outcomes, and Ownership 23.2.7 Revealing, Mitigating, and Preventing Biases in Analysis and Interpretation 23.2.8 Transparency in Feedback, and Dissemination 23.2.9 Considering Who Will Use the AI 23.2.10 Educational Opportunities for AI 23.3 Ethics and Challenges 23.4 Conclusion Acknowledgments Note References Index