Search for: All records

Abstract—Collaborative problem-based learning (PBL) in online and distributed environments presents persistent challenges related to collaboration, motivation, cognitive load, and regulation of learning processes. While advances in artificial intelligence (AI) offer new opportunities to support learners, much existing work conceptualizes AI primarily as a tool or tutor rather than as a co-regulatory partner embedded within collaborative learning systems. This paper introduces HyCoReg (Hybrid Cognition for Co-Regulated Problem Solving), a sociotechnical architecture designed to support human–AI co-regulation during collaborative complex problem solving in higher education contexts. Grounded in perspectives from complex problem solving, problem-based learning, self-regulation and co-regulation, and 4E cognition (embodied, embedded, enacted, and extended), the proposed system integrates multiple AI agents that assume complementary regulatory roles (e.g., cognitive, metacognitive, motivational, and coordination support). The architecture enables dynamic human–agent interactions through active engagement, adaptive feedback, learning analytics, and agent coordination mechanisms. This work contributes a theoretically grounded design model for hybrid cognition in collaborative problem solving and offers implications for the design of AI-supported collaborative learning environments that foreground meaningful regulation in ecological learning systems. 

Abstract—Timely, high-quality feedback is essential for learning, yet formative assessment remains difficult to scale when student work is multi-modal (typed responses, PDFs, handwritten images) and when instructors must grade across multiple activities and sections. This paper presents a reference architecture that operationalizes generative-AI grading as a workflow and captures learning evidence using Experience API (xAPI) statements for downstream analytics. The system combines a workflow-based grading agent that produces structured outputs (scores, rubric-aligned feedback, flags) and a server-side orchestration layer that retrieves submissions, executes the grading workflow, posts grades and feedback back to the learning environment, and emits xAPI records to a Learning Record Store (LRS). The architecture is designed as a research prototype to support consistent, repeatable processing rather than fully autonomous grading. The design supports assignment and quiz-style assessments, file uploads (PDF/images) and online text responses, and includes controls for idempotent processing, run logging, and safe fallback to human review. We describe deployment considerations and a study design to evaluate latency, reliability, rubric agreement with human graders, and analytics completeness in authentic course settings. 

Abstract—The integration of Large Language Models (LLMs) into aviation safety decision-making represents a significant technological advancement, yet their standalone application poses critical risks due to inherent limitations such as factual inaccuracies, hallucination, and lack of verifiability. These challenges undermine the reliability required for safety-critical environmentswhere errors can have catastrophic consequences. To address these challenges, this paper proposes a novel, end-to-end framework that synergistically combines LLMs and Knowledge Graphs (KGs) to enhance the trustworthiness of safety analytics. The framework introduces a dual-phase pipeline: it first employs LLMs to automate the construction and dynamic updating of an Aviation Safety Knowledge Graph (ASKG) from multimodal sources. It then leverages this curated KG within a Retrieval- Augmented Generation (RAG) architecture to ground, validate, and explain LLM-generated responses. The implemented system demonstrates improved accuracy and traceability over LLM-only approaches, effectively supporting complex querying and mitigating hallucination. Results confirm the framework’s capability to deliver context-aware, verifiable safety insights, addressing the stringent reliability requirements of the aviation industry. Future work will focus on enhancing relationship extraction and integrating hybrid retrieval mechanisms. 

Given the increasing reliance on UAS in sensitive applications, ensuring the confidentiality, integrity, and availability of their data streams is paramount. Traditional encryption methods often fail to balance performance and security under real-time constraints. This paper addresses this gap by proposing a hybrid adaptive encryption framework that integrates rule-based (RL) logic and machine learning (ML) to dynamically adjust encryption protocols based on data sensitivity, bandwidth, and CPU load. The experimental results demonstrate improved responsiveness and security under varied conditions using real-time simulations. The effectiveness of the system is benchmarked through execution time analysis, classification accuracy, and adaptive decision precision, highlighting its potential for secure and efficient UAS communications. 

The proposed integration of the Generalized Intelligent Framework for Tutoring (GIFT) into Curiosity Games, our Mars-based VR math game, represents a transformative approach to STEM education and training. Designed to engage middle and high school students, the game immerses learners in a dynamic educational experience while fostering skills critical for future careers in defense, aerospace, and STEM fields. Developed by the U.S. Army and its collaborators, GIFT provides advanced adaptive learning capabilities, real-time assessments, and robust monitoring tools that align seamlessly with the game’s structured classroom and exploratory open-world design. Students are able to use the game using VR or a desktop based application. The classroom component, set within a Martian observatory, will leverage GIFT to support a traditional adaptive course design. Students will follow a structured step-by-step process aligned with the 5E model: Math Conceptual Exercises (Engage and Explore), Apply Arithmetic (Explain and Elaborate), Test Questions (Evaluate), and culminating in Hands-On Activities (Elaborate). Progression is tied to performance, with students required to achieve a passing score of 80% or higher to move to the next stage. GIFT will issue credits as students successfully complete activities, tying in-game rewards to academic achievement.GIFT Integration Highlights:1. Real-Time Adaptive Support:-GIFT will provide tiered intervention to assist students who struggle with tasks, increasing teacher efficiency by prioritizing resources. These tiers include:-The AutoTutor Support is used first for immediate assistance.-Peer-to-Peer Support, where GIFT identifies proficient students to mentor peers, is the next intervention employed.-Small Group Support, where GIFT groups students with similar needs and facilitates teacher-hosted sessions, can be further employed.-One-on-One Teacher Support, dynamically pairing individual students with teachers based on real-time data to optimize outcomes, is the final intervention employed after the previous more time efficient options have been exhausted.2. Game Master Interface:-Teachers monitor progress, control adaptive exercises, and provide feedback through Objectives, Assessment, and Teams and Roles panels.3. After-Action Reviews (AAR):-Time-synced playback and video panels enable detailed reviews and targeted feedback.4. Open-World Exploration:-Students use credits to build Martian civilizations, solve tactical scenarios, and complete engineering challenges, with adaptive difficulty based on performance.Expected Outcomes:The integration of GIFT will enhance learning outcomes through personalized, adaptive support, improve teacher efficiency by optimizing resource allocation, and inspire the next generation of STEM professionals. This project aligns with the Department of Defense’s goals of fostering a technically skilled workforce and demonstrates the potential of integrating intelligent tutoring systems into immersive educational platforms.Lessons Learned:Leveraging GIFT’s existing tools minimizes development time for features such as teacher dashboards, multiplayer support, and scenario authoring. Utilizing these resources allows for efficient implementation and scalability, ensuring maximum return on investment. 

This study introduces a modular teaching framework for business data analytics (BDA) curricula and programs. The framework integrates gamification features of the SAP business processes, ERPsim Games, and SAP data warehousing into the experiential learning of BDA curricula. The pedagogical practices of deploying the framework in an undergraduate BDA course are reported and assessed in virtual and face-to-face teaching modalities. The assessment shows that integrating the framework in business pedagogies enhances the BDA learning experience and teaching effectiveness. The paper concludes with the theoretical and practical implications of the study for business educators and practitioners in BDA learning, teaching, and training. The limitations and future research avenues of the study are discussed. 

This paper presents an innovative courseware project based on the Adaptive Distributed Learning (ADL) Initiative’s Total Learning Architecture (TLA [1]), which encompasses a technical framework for education and training based on a data strategy built around open standards to support interoperability across diverse organizations and products ([2]). This framework includes definitions of a set of policies, specifications, and standards that enable a future learning ecosystem to facilitate lifelong learning principles promoting personalized and flexible learning environments that include both formal and informal activities [3]. In Fall 2023, a TLA- inspired course framework was implemented in a data visualization course for senior undergraduates and graduate students, using Moodle and a Learning Record Store (LRS) to track over 200,000 learning records. This system allowed instructors to visually monitor online learning activities for the whole class as well as selected individual learners. As future work, the course will expand to 10 STEM courses across 11 universities in the next three years as part of an existing NSF commitment. 

Over the past decade, the educational landscape has experienced a surge of online learning and instruc-tional platforms (Liu et al., 2020). This remarkable surge can be attributed to a confluence of factors, including the rising demand for higher education opportunities, the shortage of available teaching staff, and the rapid advancements in information technology and artificial intelligence capabilities. Artificial Intelligence (AI) remained a niche area of research with limited practical applications in education for over half a century (Bhutoria, 2022; Chen et al., 2020; Roll & Wylie, 2016) from 1950 to 2010. Howev-er, in recent years, the advent of Big Data and advancements in computing power have propelled AI into the educational mainstream (Alam, 2021; Chen et al., 2020; Hwang et al., 2020). Today, the rise of machine learning, deep learning, automation, together with advances in big data analysis has sparked novel perspectives and explorations around the potential of enhancing personalized learning, a long-term educational vision of technology-enhanced course options to meet student needs (Grant & Basye, 2014). Fostering personalized learning necessitates the development of digital learning environments that dynamically adapt to individual learners' knowledge, prior experiences, and interests, while effectively and efficiently guiding them towards achieving desired learning outcomes (Spector, 2014, 2016). AI-powered technologies have made it possible to analyze data generated by learners and provide instruc-tion that matches their learning performance. Through learning analytics and data mining techniques, large datasets collected are analyzed and processed to uncover learners' unique learning characteristics, often referred to as learner profiling (Tzouveli et al., 2008). Subsequently, leveraging artificial intelli-gence algorithms, the learning content is tailored, and personalized learning paths are designed to align with each learner's identified needs and preferences, thereby facilitating personalized learning experienc-es. 

Artificial intelligence (AI) leverages mathematical algorithms to emulate human cognitive abilities, leading to a transformative impact on the education sector. Educators are at the front lines of implementing AI in the classroom. Recent scientific studies demonstrate the capacity of AI, particularly generative models like ChatGPT, to reshape various aspects of education. In a recent study, we showcased that the integration of both artificial intelligence, specifically ChatGPT, and interactive learning activities significantly enhances the engagement levels of STEM students enrolled in a General Biology course. Furthermore, this combined approach not only boosts student engagement but also demonstrates an improvement in their overall performance within the course. Building on preliminary studies, the objective of this review article is to delineate the diverse applications of generative AI in education. To achieve this objective, we conducted a thorough search across scientific databases, including Google Scholar, Science Direct, government websites, and other resources, to collect relevant papers. Our findings underscore the contributions of generative AI, exemplified by ChatGPT, in enabling students to generate innovative text for written assignments, providing personalized feedback, facilitating adaptive learning, enhancing accessibility to education by eliminating barriers for individuals with disabilities, and supporting research endeavors. 

The current National Airspace System (NAS) is reaching capacity due to increased air traffic, and is based on outdated pre-tactical planning. This study proposes a more dynamic airspace configuration (DAC) approach that could increase throughput and accommodate fluctuating traffic, ideal for emergencies. The proposed approach constructs the airspace as a constraints-embedded graph, compresses its dimensions, and applies a spectral clustering-enabled adaptive algorithm to generate collaborative airport groups and evenly distribute workloads among them. Under various traffic conditions, our experiments demonstrate a 50% reduction in workload imbalances. This research could ultimately form the basis for a recommendation system for optimized airspace configuration. Code available at https://github.com/KeFenge2022/GraphDAC.git