generative ai

Predicting Teachers' Behavioral Intention to Adopt Generative AI in Teaching: An Integrated TAM-UTAUT Regression Model

Generative artificial intelligence (GenAI) has entered classrooms faster than most institutions have been able to formulate policy, yet its instructional value ultimately depends on whether teachers choose to use it. This study examined the determinants of teachers' behavioral intention (BI) to use GenAI in teaching, drawing on an integrated Technology Acceptance Model (TAM) and Unified Theory of Acceptance and Use of Technology (UTAUT) framework. Five predictors were specified: perceived ease of use (PEoU), perceived usefulness (PU), social influence (SI), facilitating conditions (FC), and anxiety (ANX). A cross-sectional survey was administered to 500 in-service teachers drawn from three Chinese educational institutions spanning medical higher education, vocational higher education, and primary education. Each construct was operationalized as a composite mean of its constituent items, and the model was estimated using the Regression module of SmartPLS 4 with bootstrapping (5,000 subsamples) to obtain confidence intervals. Collinearity diagnostics were acceptable (VIF = 1.046-1.238). The model accounted for 32.7% of the variance in behavioral intention (adjusted R² = .320). Perceived usefulness was the strongest predictor (β = .228, p < .001), followed by facilitating conditions (β = .206, p < .001), perceived ease of use (β = .198, p < .001), and social influence (β = .195, p < .001). Contrary to expectations, anxiety exerted no significant effect (β = .018, p = .631), with a bootstrap confidence interval that spanned zero. The findings indicate that teachers' adoption decisions are jointly driven by instrumental value and institutional provisioning, and that generalized technology anxiety is not, in itself, a barrier among teachers who already have practical exposure to GenAI. Implications for professional development design and institutional AI policy are discussed.

Artificial intelligence in healthcare: Technical advances, clinical integration, and future perspectives

This narrative review examines the evolution of artificial intelligence (AI) in healthcare, with a focus on the transition from early rule-based systems to modern deep learning architectures and their integration into clinical practice. We examine foundational technologies, including convolutional neural networks for image interpretation, vision transformers for modeling long-range dependencies, and generative adversarial networks for image reconstruction and synthesis. The review further discusses the emergence of multimodal foundation models that integrate imaging with textual and genomic data to enhance diagnostic robustness. The application of these technologies is analyzed across three primary domains: Radiology (image enhancement and automated interpretation), cardiology (electrocardiographic and echocardiography analysis), and oncology (tumor classification and treatment planning). Specific attention is given to the national context in Türkiye, highlighting local initiatives such as TEKNOFEST and TÜBİTAK-supported projects that foster domestic AI development. While AI offers significant benefits in terms of diagnostic accuracy and treatment workflow optimization, challenges regarding data privacy, algorithmic bias, and interpretability (“black box” issues) persist. Future progress depends on the development of explainable AI, rigorous prospective validation, and the establishment of ethical regulatory frameworks.

The Effects of Self-Learning Generative AI NPCs on Player Game Immersion and Loyalty: The Mediating Role of Game Immersion

With the growing use of generative artificial intelligence in digital games, whether AI-driven non-player characters (NPCs) can improve player experience and further shape continued play intention has become an important question for game design and user research. This study examines whether self-learning generative AI NPCs influence game loyalty through game immersion. A quasi-experimental between-subjects design and a retrospective questionnaire were used to compare players who had interacted with self-learning generative AI NPCs in the past three months with players who had only interacted with traditional scripted NPCs. Based on 195 valid responses, the AI NPC group reported significantly higher game immersion and game loyalty than the traditional NPC group. Mediation analysis using PROCESS Model 4 further showed that game immersion significantly mediated the relationship between NPC type and game loyalty. The findings suggest that the value of generative AI NPCs lies not only in the technology itself, but also in whether adaptive and natural interaction can enhance immersion and thereby relate to continued play and recommendation intention. Given the non-random grouping and self-reported data, the findings should be interpreted cautiously.

Algorithmic Publicity and the Right to a Fair Trial

This article examines how social media and AI-driven algorithmic publicity affect the right to a fair trial in Australian criminal jury proceedings. It outlines the common law and statutory controls on prejudicial publicity, including sub judice contempt, permanent stays, jury directions and evidentiary discretions, and shows how they operate in leading High Court and appellate decisions and recent law reform work. It then explains how search engines, social media feeds, recommender systems, synthetic media and generative AI, through engagement-optimised ranking and amplification, undermine broadcast-era assumptions about juror exposure. Drawing on empirical research on juror psychology, media effects and juror internet use, the article evaluates the practical capacity of safeguards such as contempt, stays (including delay, change of venue and sequestration), judge-alone trials and directions where prejudicial material is persistent, searchable and personalised. It argues for a robust fair-trial standard paired with a more candid account of digital exposure. Building on developments in the UK, EU, US, China and Estonia, it proposes doctrinal and procedural refinements, narrowly targeted platform duties and court-supervised AI monitoring tools to preserve jury trial and open justice while maintaining credible fair-trial guarantees in an algorithmic information environment.

Ethically integrated generative AI for reading and vocabulary development in higher education: an experimental study of efficacy and learner perceptions

The study fills a gap in the field of applied linguistics, which has not been widely explored, by considering the pedagogical effectiveness and ethical use of AI in higher education. Addressing an underexplored empirical area in applied linguistics, the research builds on recent studies such as Yin and Hanif (2024) and focuses on pedagogical effectiveness and ethical implementation in higher education. A pretest-posttest control group design was used with 148 first-semester undergraduate students. The control group (n = 74) received no intervention, while the experimental group (n = 74) was assigned to a 15-week program using AI via ChatGPT, Claude AI, Meta AI, and Gemini, along with structured ethical practices, including evaluating AI output, paraphrasing, and transparency. The control group was taught in a traditional manner (n = 74). The quantitative results indicated that the AI group had significantly higher scores in both vocabulary and reading comprehension (p < .001). Immediate improvements in vocabulary were observed during the early testing period, but leveled off in the control group in later tests; reading comprehension took time to develop but was maintained. There were also gender differences, possibly due to differences in learning preferences, as males showed greater vocabulary growth whilst females showed greater reading comprehension. The qualitative findings from interviews and surveys based on the TAM indicated that the use of AI would provide greater learner autonomy, reduce cognitive load for learners, and be highly motivated to continue using AI, with ethical usage as a key factor. The results showed that the use of AI in education, guided by ethical principles, can effectively enhance learning outcomes, as evidenced by a 30% improvement in scores related to learning autonomy. The study provides a replicable model for the ethical use of AI in higher education settings in the Global South, which has been applied in Pakistan and can be adapted for other regions. It suggests training faculty to use ethical AI in the classroom through workshops on paraphrase protocols and bias detection, providing better access to AI and infrastructure, and conducting longitudinal studies over 6–12 months following the intervention to measure sustained learning outcomes and retention loss.

Developing an LLM-Based Feedback System Grounded in Evidence-Centered Design to Support Physics Problem Solving

Generative AI offers new opportunities for individualized and adaptive learning, e.g., through large language model (LLM)-based feedback systems. While LLMs can produce factually correct feedback for relatively straightforward conceptual tasks, delivering high-quality feedback for tasks that require advanced domain expertise—such as physics problem solving—remains a substantial challenge. This study presents the design and implementation of an LLM-based feedback system for physics problem solving grounded in evidence-centered design and reports a first evaluation within the German Physics Olympiad. Participants rated the usefulness and correctness of the generated feedback for each implemented problem. The collected ratings indicate that the feedback was generally perceived as useful and highly correct. However, an in-depth analysis revealed that the feedback contained errors in 20% of cases—errors that often went unnoticed by the students. We discuss the risks associated with uncritical reliance on LLM-based feedback and outline potential directions for generating more adaptive and reliable LLM-based feedback in the future.

A Study on the Application and Intervention Model of Generative AI-Based Art Therapy - Focusing on the “Anxin AI” Drawing-Based Psychological Assessment System -

Artificial Intelligence in Adaptive Learning for Education: A Bibliometric Analysis

Artificial intelligence (AI) has emerged as a transformative technology in education, particularly adaptive learning, which supports personalized and learner-centered experiences. Despite the rapid growth of research on AI for adaptive learning, a comprehensive understanding of its publication performance, thematic structure, and technological evolution remains limited. To overcome this gap, the study aims to systematically map and evaluate the development of research on AI for adaptive learning in education using bibliometric methods. A bibliometric analysis was conducted using data from the Scopus database covering the period from 2015 to 2026. The study analyzed 2,354 publications through two complementary methods: performance analysis and science mapping. Performance analysis was used to evaluate publication growth, influential countries, sources, and highly cited documents. Science mapping techniques, including keyword co-occurrence analysis, were used to identify major research themes and the emerging AI technologies landscape. The findings show that research output has significantly risen after 2023 and recorded the highest number of publications in 2025. India emerged as the most productive country, while the United States had the highest citation impact. Citation analysis highlights adaptive and personalized learning systems, intelligent educational systems, and generative AI applications as key intellectual foundations of the field. Science mapping further revealed that machine learning, intelligent tutoring systems, generative AI, learning analytics, and large language models are central AI technological themes in adaptive learning research. These recent trends indicate a growing shift towards conversational AI, generative AI, and personalized intelligent learning environments. In conclusion, this study provides insights into publication trends, contributors, research themes, and emerging AI technologies in adaptive learning, which could assist researchers, educators, policymakers, and educational technology developers in improving intelligent adaptive learning systems.

Ethical Issues and Researcher Responsibility in the Use of Generative AI in Beauty Research

This study employed a systematic literature review to examine research trends in artificial intelligence within the field of beauty and cosmetology and to explore the ethical issues associated with the use of generative AI in academic research. A total of 135 studies were analyzed. The findings indicate that AI-related research in beauty and cosmetology has shown steady growth since 2020, with a particularly sharp increase between 2024 and 2025. Despite this rapid expansion, only one study explicitly addressed ethical issues, revealing a substantial gap between technological advancement and ethical reflection in the field. The analysis identified several major ethical concerns related to the use of generative AI, including copyright and authorship issues involving AI-generated images and textual content, as well as risks to personal data protection during the collection, processing, and utilization of research data. Furthermore, this study argues that beauty research represents a unique domain in which bodily data and aesthetic norms are closely intertwined, making it difficult to address its ethical challenges through general AI ethics frameworks alone. The findings further suggest that existing AI ethical principles are insufficient to address the distinctive complexities associated with the use of generative AI in beauty research. Accordingly, this study emphasizes the need for systematic regulatory frameworks, standardized disclosure guidelines, and strengthened researcher responsibility to ensure transparency, accountability, and ethical integrity in AI-assisted research practices. In particular, this study contributes to the literature by identifying domain-specific ethical issues in beauty research and highlighting the need for a specialized ethical framework governing the responsible use of generative AI.

An Overview of GenAI 2.0 Partnering With Digital Twin to Enhance Decision Making

Generative AI (GenAI) has moved on to a “2.0” level, where capability comes not only from better generators but from integrating generators with retrieval, tool use, and orchestration. The approach taken in this overview is a systems perspective, where a brief typology of families and of models is codified, along with a distillation of design patterns on how bare generators can be refined into reliable systems, namely containerized RAG, tool schemas, eventdriven context, and conservative agent systems. In this position, we propose a practice-driven assessment perspective integrating task quality, along with Attribution, Robustness with respect to distribution shift & adversarial prompts, safety & governance (Privacy, IP, Provenance), and End-to-End Efficiency in Latency, Cost, & Energy. Text/Code, Vision-Language, Speech/Audio, Networks, Public Services, & Critical Infrastructure applications provide examples of these trends extending from benchmarking. The issues it brings to light include controllability and verifiability for long-range, multiple tool agents, scalable contamination aware assessment, traceability preserving data and model stewardship, as well as efficiency for resource-constrained and edge scenarios, while pointing towards near-term areas such as tool interface interoperability, traceability aware retrieval, lifecycle visibility, reporting card interoperability, and reliability assessment. Focusing on operational criteria, the research has contributed to creating a impact for making GenAI 2.0 Trustable AI. The blueprint can be utilized as a guide for future research.

Can Large Language Models Serve as Data Analysts? A Multi-Agent Assisted Approach for Qualitative Data Analysis

Context: Manual qualitative data analysis is time-intensive and can compromise validity and replicability, affecting analysis design, implementation, and reporting. Large Language Models (LLMs) enable human-bot collaboration in Software Engineering (SE), but their potential for qualitative data analysis in SE remains largely unexplored. Objective: The objective of this study is to design and develop an LLM-based multi-agent system that synergizes human decision support with AI to automate various qualitative data analysis approaches. Methods: We used LLM-based multi-agents systems to assist the qualitative data analysis process, deploying 27 agents, each responsible for a specific task, such as text summarization, initial code generation, and extracting themes and patterns. Results: The main findings are: (1) the LLM-based multi-agent system accelerates the qualitative data analysis process, (2) the system effectively automates tasks such as text summarization, initial code generation, and theme extraction, and (3) the publicly accessible code facilitates validation and further evaluation. Conclusion: The proposed LLM-based multi-agent system automates qualitative data analysis process, creating opportunities for researchers and practitioners. Future improvements focus on enhancing multilingual performance and integrating continuous expert feedback. The source code of proposed system and system details can be found here: https://github.com/GPT-Laboratory/Qualitative-Analysis-with-an-LLM-Based-Agentts

Engineering a Governance-Aware AI Sandbox: Design, Implementation, and Lessons Learned

Collaborative AI experimentation in industry-academia requires environments that support rapid trials while maintaining controlled access, organisational isolation, and traceable workflows. Although interest in AI sandboxes is increasing, practical guidance on designing and building governance-aware experimentation platforms remains limited. This work designs and operationalizes a governance-aware, multi-tenant AI sandbox that supports structured experimentation and produces reusable evaluation evidence across stakeholders. The sandbox was developed in an industry-academia ecosystem using iteratively validated requirements gathered from industrial partners. The solution adopts a layered reference architecture that separates a multi-tenant presentation layer from a backend control plane and isolates execution and data management concerns into dedicated layers. The sandbox supports governed onboarding, project-based collaboration, controlled access to AI services, and traceable experimentation through approval workflows and audit logging. By structuring experiment context and governance decisions as persistent records, the sandbox enables evaluation evidence to be reused and compared across projects and stakeholders. The development experience yields lessons learned and practical considerations that inform deployment and future evolution of governance-aware sandbox platforms.