Search Results (769)

In complex environments, Operators need to manage continuous, real-time information flows while handling unexpected situations within a limited timeframe. This can lead to high cognitive load, stress, fatigue, etc. To prevent such situations, Artificial Intelligence (AI) systems are increasingly being considered. Their role is no limited to assistance, but extends to performing some or all of the tasks initially carried out by the operator. However, inappropriate allocation of tasks between humans and machines can exclude the operator from the loop or reduce their vigilance. This paper proposes the design and implementation of three strategies for the dynamic reallocation of tasks between a human and an AI, considering factors related to the operator (cognitive load, stress) and their activity (activity modeling). An evaluation is conducted to compare three strategies. The first two are hybrid strategies, in which both the operator and the AI can modify task allocation. The first hybrid strategy is based on self-assessment, while the second is based on activity modeling. The third strategy is an adaptable strategy, in which only the operator can change task allocation. The use case is an emergency call center simulation implemented on InteractiveAI Preliminary findings from our user exploratory study suggest that participants tended to better accept adaptable automation, while also exhibiting a higher error distribution compared to hybrid automation strategies. No significant differences were observed in cognitive load or situational awareness in this limited sample. However, recurring instances of mode confusion were observed with hybrid strategies. Full article

Large language models (LLMs) are increasingly being used in education, yet their correctness alone does not capture the quality, reliability, or pedagogical validity of their problem-solving behavior, especially in mathematics, where multi-step logic, symbolic reasoning, and conceptual clarity are critical. Conventional evaluation methods largely focus on final answer accuracy and overlook the reasoning process. To address this gap, we introduce a novel interpretability framework for analyzing LLM-generated solutions using undergraduate calculus problems as a representative domain. Our approach combines reasoning flow extraction and decomposing solutions into semantically labeled operations and concepts with prompt ablation analysis to assess input salience and output stability. Using structured metrics such as reasoning complexity, phrase sensitivity, and robustness, we evaluated the model behavior on real Calculus I–III university exams and compared it with the performances of students enrolled in the courses. Our findings revealed that LLMs often produce syntactically fluent yet conceptually flawed solutions with reasoning patterns sensitive to prompt phrasing and input variation. This framework enables a fine-grained diagnosis of reasoning failures, supports curriculum alignment, and informs the design of interpretable AI-assisted feedback tools. The framework was evaluated on Gemma 3, an open-access large language model, across zero-shot, retrieval-augmented generation, and contextual retrieval configurations, using nine real undergraduate calculus examinations from three course levels. To our knowledge, this is the first paper to apply a combined reasoning flow decomposition and prompt ablation framework to real undergraduate calculus examinations, benchmarked against actual student cohort performance, laying the foundation for the transparent and responsible deployment of AI in STEM learning environments. Full article

Transitioning toward efficient and digital industrial design requires preliminary tools that support early decision-making in plant layout studies. This qualitative and exploratory study analyzes an Artificial Intelligence (AI)-assisted and Augmented Reality (AR)-supported workflow within the Systematic Layout Planning (SLP) framework for the preliminary spatial evaluation of a cosmetic emulsion production plant. The study was developed as a case study based on a previously reported layout for obtaining cosmetic emulsions from Amazonian oils. A top-view layout was examined through structured prompts aligned with SLP criteria, including product journey, activity relationships, relational diagrams, and space requirements. ChatGPT was used only as a qualitative reasoning assistant, without optimization, prediction, mathematical modeling, or algorithmic functions. After the AI-assisted review, the refined layout was represented in three dimensions and visualized through AR in a real environment. The results identified potential improvements related to operational flow, traceability, critical area relationships, and spatial organization. AR-assisted visualization provided preliminary visual evidence of compatibility between the refined layout and the selected site, supporting an early review of circulation, access, and volumetric behavior. The sequential integration of SLP, AI, and AR is proposed as an exploratory workflow for early-stage layout evaluation, pending future quantitative validation studies and expert technical review. Full article

Bipolar plates are critical components of proton exchange membrane fuel cells (PEMFCs), strongly influencing thermal management, mechanical stability, and overall system efficiency. In this study, an integrated framework combining multiphysics simulation, artificial intelligence (AI), and data augmentation techniques was developed for PEMFC bipolar plate optimization. A coupled thermal–structural finite element model was established in COMSOL Multiphysics to evaluate temperature distribution, thermal stress, and structural deformation under varying operating conditions. A total of 80 parametric design cases were generated by varying six key parameters: hole radius, plate thickness, heating power, manifold pressure, plate number, and heat transfer coefficient. The dataset was expanded using SMOTE, GAN, and LLM-based augmentation techniques and used to train ANN, LR, RF, XGBoost, and SVR models. Model performance was evaluated using 5-fold cross-validation with MAE, RMSE, and LogCosh metrics. The results showed that ensemble tree-based methods, particularly RF and XGBoost, achieved the highest prediction accuracy and computational efficiency. XGBoost produced the best temperature prediction performance for the SMOTE-based dataset (RMSE = 3.668), while RF achieved the lowest stress prediction error (RMSE = 0.0490). GAN-augmented datasets provided stable and reliable predictions, whereas LLM-generated datasets resulted in higher prediction errors and lower physical consistency. Feature importance analysis revealed that plate thickness dominates displacement prediction (≈0.72 importance), manifold pressure governs stress behavior (≈0.999), and heating power is the primary factor affecting temperature prediction. The proposed AI-assisted surrogate modeling framework enables rapid and accurate thermo-mechanical prediction while significantly reducing computational cost compared to conventional multiphysics simulations. The findings demonstrate that integrating physics-based simulations with data-driven approaches provides an efficient strategy for the optimization of next-generation PEM fuel cell bipolar plates. Full article

Global salinization affects approximately one billion hectares of land in more than 100 countries, posing a severe threat to food security and ecosystem sustainability. Microbial remediation using plant growth-promoting microorganisms offers an eco-friendly alternative to physicochemical methods. However, bridging the gap between laboratory cultivation of single strains and field-scale application of synthetic microbial communities (SynComs) remains difficult, owing to inconsistent efficacy and a lack of unified design frameworks. This review examines the evolution from single strains to rationally designed SynComs for saline soil remediation. A ‘structure–function–mechanism’ framework is proposed, integrating five core microbial modules, namely ion regulation and osmotic stabilization, ethylene and phytohormone modulation, antioxidant activation, nutrient cycle activation, and systemic resistance induction. The review elucidates key determinants of synthetic community success, including functional complementarity, strain compatibility, and host–environment matching, while revealing a marked quantitative gap between controlled experiments and field performance. Key bottlenecks are identified, including the lack of high-throughput compatibility screening, poorly quantified long-term ecological risks, and the absence of standardized application guidelines across agro-ecological zones. Finally, emerging avenues are discussed, such as microbial–microalgal symbiosis and AI-assisted design, outlining a roadmap for next-generation smart microbial products integrated into climate-resilient farming systems. Full article

Internet of Things (IoT), sensor-network, and cyber-physical system (CPS) software increasingly relies on large language models (LLMs) and autonomous agents for code generation, maintenance, and vulnerability repair. However, LLM-generated edge services, telemetry APIs, configuration handlers, and data-aggregation routines can introduce SQL injection, path traversal, command injection, hard-coded credentials, and unsafe device-control logic, which may compromise sensing data integrity and system safety. Existing approaches largely rely on static post hoc analysis and lack a unified modeling of the generation process, making it difficult to achieve a principled trade-off between functionality and security. To address this challenge, we propose SafeCodeRL, a framework that integrates multi-agent collaboration with constrained reinforcement learning for trustworthy LLM-generated IoT/CPS software. SafeCodeRL models code generation as a security-aware sequential decision process, where Planner, Code, Security, Test, and Critic agents jointly optimize task decomposition, code synthesis, vulnerability auditing, and sandbox-based validation. We design a constraint-aware policy based on Proximal Policy Optimization, augmented with a Lagrangian mechanism and a shielding strategy to explicitly enforce security constraints. Experiments on real-world engineering and security benchmarks, including SWE-bench, SecurityEval, and CyberSecEval, show that SafeCodeRL reduces high-risk vulnerabilities by over 60% while maintaining high functional correctness. A scenario-level IoT/CPS case study further demonstrates that SafeCodeRL substantially improves secure pass rates for sensor telemetry, edge gateway, configuration-management, and data-aggregation tasks, providing a practical path toward trustworthy AI-assisted software development for sensor-driven systems. Full article

This narrative review examines the historical evolution, current applications, and major challenges of machine learning (ML) in education, positioning ML as a transformative yet deeply contested force in contemporary teaching and learning. Tracing developments from early computer-assisted instruction and intelligent tutoring systems to contemporary deep learning, natural language processing, and generative AI, the review shows how these technologies have expanded education’s capacity for personalization, prediction, automation, content generation, and large-scale data-driven decision-making. It synthesizes evidence across key domains, including student performance prediction, early warning systems, adaptive learning, intelligent tutoring, automated assessment, learning analytics, curriculum design, and inclusive education. In addition, the review critically highlights persistent limitations and risks, particularly algorithmic bias, data privacy concerns, limited interpretability, uneven pedagogical value, infrastructure constraints, and the disruption of conventional assessment by generative AI. Rather than treating ML as a purely technical innovation, the paper argues that its educational significance depends on how responsibly it is designed, implemented, and governed. The review concludes that the future of ML in education will be shaped not only by advances in computational methods but also by ethical judgment, pedagogical alignment, and institutional commitment to equity, transparency, and human-centered educational practice across diverse learning contexts worldwide. Full article

Architectural, urban design, and planning practices are increasingly expected to demonstrate measurable impact, accountability, and responsiveness to complex environmental and social challenges. Evidence-based design (EBD) and research-informed design (RID), which ground design decisions in systematically gathered and critically evaluated knowledge, offer a structured pathway to bridge research and practice. Despite growing recognition, however, EBD and RID remain unevenly integrated across professional practice, particularly within small and medium-sized enterprises (SMEs), which constitute the majority of firms in Europe. This paper explores how SMEs understand, adopt, and operationalize research within architectural, urban design, and planning processes, while identifying the factors that enable or constrain the integration of research into practice. Drawing on a qualitative multiple-case study of four European firms located in Cyprus, Portugal, Italy, and Croatia the study uses semi-structured interviews and thematic analysis supported by AI-assisted coding to identify patterns in how systematic research is understood, enacted and positioned in everyday SME practices. The findings show that research integration depends less on firm size than on the interplay between client expectations, organizational culture, and professional ideology. Practices span a spectrum ranging from ad hoc, compliance-oriented, and project-specific inquiry to strategically embedded and, in one case, activist research-led modes. While research engagement can enhance credibility, efficiency, and innovation, persistent barriers—including limited resources, client resistance, deficient knowledge-management routines, and the absence of shared evaluative frameworks—continue to hinder systematic adoption. Building on the cross-case analysis, the paper proposes a conceptual framework of different modes of research integration in SMEs, offering a heuristic lens for understanding how organizational and contextual factors shape the uptake of research in design practice. The findings contribute to ongoing discussions on practice-based research and highlight the need for more context-sensitive approaches to research integration in small and medium-sized design firms. Full article

Literature reviews are essential for synthesizing existing knowledge, mapping research domains, identifying intellectual structures, and highlighting research gaps within a field. However, many literature reviews are incomplete because database search strategies are not adequately specified or validated. Search strategies are frequently underreported and undermotivated across the systematic review literature and bibliometrics, while query formulation remains time-consuming, error-prone, and particularly difficult in interdisciplinary or rapidly evolving topics. This article fills that void by developing a guideline for designing a professional topic query in existing academic databases and emphasizing search design as the front-end validity problem in bibliometric research. The article uses the Artificial Intelligence Think Tank framework as a methodological engine and applies it to bibliometric retrieval engineering via structured interaction with generative AI systems and human experts. The paper assists scholars performing bibliometric studies, scientometric analyses, systematic literature reviews, scoping reviews, and hybrid evidence-synthesis projects. Full article

Effective dementia care is often hindered by fragmented communication among patients, informal caregivers, and clinicians. To address this, we introduce an ambient assisted living (AAL) framework designed to establish a continuous, virtual, and unobtrusive connection between an elder’s home and external guardians or medical staff (virtual rounds). The system enables guardians to communicate directly within the home environment, without requiring the older adult to manually accept calls or activate the connection using wearable devices, buttons, or other interfaces. The elders can activate the connection verbally. The structural core of this system relies on three novel hardware configurations designed for zero-interface operation: a remote audio announcement device, a bidirectional intercom, and a “zero-interface mirror” enabling stream-only, real-time video co-presence between patients and guardians. Crucially, the system utilizes a privacy-preserving, staged edge-AI architecture to process data. By default, it operates without long-term persistent storage, selectively transmitting abstracted audio-based behavioral metrics to a secure dashboard. For advanced dementia stages, the system employs ephemeral data retention—specifically a highly restricted, 24 h rolling audio buffer—allowing authorized guardians to verify acute events without permanently exfiltrating raw data. We evaluate this infrastructure through a 10-month longitudinal, single-home feasibility deployment, augmented with historical verified fall data to rigorously test the detection of rare acute events. The study validates the framework’s technical viability, system uptime, and privacy-first architecture in continuously tracking long-term proxy behavioral indicators under real-world conditions. Rather than asserting generalized clinical efficacy, this work demonstrates the operational feasibility of a novel, affordable, technical blueprint for dignified, remote digital care coordination. Full article

Evaluating AI-assisted tools in healthcare human–computer interaction (HCI) presents methodological challenges when practical constraints limit sample sizes. Standard pooled statistical analysis can then produce misleading results, including Simpson’s Paradox, where aggregate trends contradict patterns observed within subgroups. This paper introduces a conditional Gaussian model framework that models each experimental condition separately rather than pooling all observations. Through a within-subjects evaluation of an AI-assisted UI/UX design tool for medical software interfaces (n = 4 professional designers), we demonstrate how pooled analysis produced a misleading negative correlation between design time and IEC 62366 compliance (the medical device usability standard; pooled $r=-0.76$, $p=0.029$, $n=8$), even though every designer achieved both faster times and higher compliance with the AI tool. Within-condition correlations were non-significant and inconsistent in sign, confirming the pooled association as an aggregation artefact rather than a within-designer trade-off. The conditional analysis surfaces experience-indexed differences: the less UI-experienced designer showed the largest time reduction (up to 92%), while the two high-AI-experience designers showed the largest automated proxy-compliance gains (+25 to +29 percentage points). Sample standard deviations were also lower in the AI-assisted condition than in the traditional condition for both outcomes (time: $20.0\to 11.3$ min; compliance: $10.6\to 7.6$ percentage points); at $n=4$ per condition, however, this difference in variance can neither be confirmed nor falsified, and we make no inferential claim about variance compression. A follow-up phase (n = 3) that adapted the tool’s scaffolding to designer experience yielded a bidirectional response, with the two high-AI-experience designers further reducing time and the less UI-experienced designer engaging more deeply with the design output. Because all participants completed the traditional condition before the AI-assisted condition, the study is interpreted as a sequentially unbalanced exploratory comparison, not as a counterbalanced causal test of tool effectiveness. We provide guidelines for healthcare HCI researchers facing sample-size constraints endemic to specialised domains. Full article

Background: Prenatal diagnosis of life-limiting fetal conditions often leads to counseling focused primarily on therapeutic abortion. Perinatal hospice has emerged as an alternative model of care for families who choose to continue the pregnancy. This paper has two primary aims. First, it discusses structured perinatal hospice programs and their role in supporting parental decision-making after such diagnoses, with attention to ethical and emotional complexities. Second, the paper introduces NOVA-L (Navigating Options & Vital Assistance for Life-limiting conditions), a conceptual Decision Support System (DSS) designed to complement perinatal hospice care. Methods: The paper provides a conceptual and descriptive analysis of the Comfort Care clinical model. It also outlines the proposed architecture of NOVA-L. DSSs combine clinical guidelines, research data, and outcome registries on digital platforms, providing evidence-based information and AI-supported analytical tools. Their potential adaptation to perinatal hospice care is explored. Results: The Comfort Care model involves interdisciplinary counseling, structured communication, and psychosocial support to facilitate clarification of parental values and care pathways. NOVA-L is presented as a complementary tool that may enhance transparency in risk evaluation and option comparison through accessible interfaces under professional supervision. Conclusions: Structured perinatal hospice programs may enhance clarity and compassion in decision-making. The conceptual integration of AI-supported DSS tools, such as NOVA-L, could strengthen ethically grounded, emotionally sensitive parental support. Full article

The NPRR research project “From waste to manufacturing” developed an AI-powered digital tool to support the transition towards a circular built environment in Italy. The tool integrates a web platform enabling data exchange about materials with recycled content between designers, manufacturers and waste recyclers, with a CAD plug-in for real-time sustainability assessment. As such, the tool fosters the use of recycled materials and allows a reduction in embodied energy. AI, trained through scenario-based learning and stakeholder participation, assists designers in sourcing recycled materials and processing data. With further training by LCA experts, it could interpret Environmental Product Declarations to guide material selection in line with international regulations. Full article

Metasurfaces exploit subwavelength scattering elements to manipulate light with a level of flexibility that is difficult to achieve using conventional optical platforms, making them promising building blocks for next-generation photonic systems. Yet the increasing dimensionality of metasurface design spaces and the demand for multifunctional responses have exposed the limitations of traditional intuition-led design approaches. In this Review, we survey the emergence of artificial intelligence (AI)-empowered metasurfaces across three major themes: inverse design, multidimensional optical-field control, and free-space optical computing. We first summarize the fundamental principle of optical field manipulation and the algorithmic approaches to metasurface design, including stochastic optimization, deep neural networks, and generative models, with emphasis on their capabilities in rapid performance prediction and inverse structural discovery. We next discuss artificial intelligence-assisted strategies for engineering multiple spatial, spectral, and polarization degrees of freedom in free space. We then highlight the role of AI-empowered metasurface architectures in optical information processing and computation. Together, these developments point to a powerful framework for integrating machine intelligence with meta-optics, with implications for autonomous photonic systems and high-capacity optical computing. Full article

Since many children with Down syndrome have difficulties with emotion recognition, there is a significant application gap in assistive technologies and affective computing that could be addressed. Conventional deep learning methods, which depend on the standard raster-scan flattening operation, achieve limited accuracy in this population because they fail to preserve spatial locality. In this paper, we propose a novel Hilbert space-filling curve optimization for neural network flattening layers, specifically designed not only to address these gaps in assistive technologies for this vulnerable group who are currently underserved by affective computing, but also to provide a framework for researchers seeking to fine-tune the architecture of artificial neural networks. Our approach retains spatial coherence using Hilbert indexing, implemented as flexible Keraslayers that are compatible with standard architectures such as VGG16 and ResNet50. A comprehensive analysis across multiple datasets reveals a 4% improvement in emotion recognition accuracy compared to Hilbert. The Hilbert optimization achieves 71% precision in Down syndrome emotion classification while reducing processing overhead by approximately 5%. By closing the emotion recognition gap with spatial-aware deep learning, our work contributes to more equitable AI for healthcare and advances the development of assistive technologies for neurodiverse populations, with near-term clinical utility in pediatrics and broader applications in affective computing. Full article