This study investigates the transformative potential of AI-enhanced e-learning platforms in empowering religious leaders, such as Imams and Muazzins, to drive sustainable agricultural practices in rural Bangladesh. By integrating artificial intelligence-driven tools with digital education, the research equips these community influencers with essential agricultural knowledge, digital literacy, and entrepreneurial skills, enabling them to mentor small-scale farmers in areas like pisciculture, poultry farming, and crop cultivation. The proposed framework addresses key challenges in rural economies, including low productivity, limited market access, and vulnerability to climate variability, while fostering economic resilience and reducing poverty through optimized resource management and direct market linkages. Drawing on mixed-methods analysis, including surveys, interviews, and pilot case studies, the model highlights how AI applications—such as predictive analytics for weather forecasting, pest detection via image recognition, and yield optimization—can enhance farming efficiency by 15–30%, as evidenced by recent initiatives in South Asia. Barriers like inadequate infrastructure, digital divides, and cultural resistance are mitigated through partnerships with NGOs, government bodies, and tech firms, alongside culturally aligned content rooted in Islamic principles of stewardship. The findings underscore the role of religious leaders as catalysts for technological adoption, bridging traditional communities with modern innovations to promote socio-economic equity and long-term sustainability in Bangladesh's agricultural sector.
Large Language Models (LLMs) exhibit impressive generative capabilities but remain prone to hallucinations — plausible yet false statements produced with high confidence. Such phenomena undermine trust and reliability in sensitive domains including health, law, and cybersecurity. Despite significant progress in retrieval-augmented generation (RAG), calibration methods, and mixture-of-experts (MoE) architectures, existing systems still lack a unified framework for veridiction, abstention, and energy-aware reasoning. This work introduces S-AI Against Hallucinations, a bio-inspired and parsimonious architecture derived from the Sparse Artificial Intelligence (S-AI) framework. The proposed model implements a symbolic-hormonal orchestration mechanism that enables generative agents to detect uncertainty, abstain when appropriate, and maintain citation integrity under ambiguous or adversarial conditions. The system relies on four hormonal variables — Hallucination Uncertainty (HU), Citation Integrity (CI), Contradiction Observer (CO), and Retrieval Evidence (RE) — dynamically regulated by a MetaAgent through hysteresisbased thresholds. Experiments performed on diverse scenarios, including factual question answering, scientific summarization, numerical reasoning, and out-of-distribution prompts, demonstrate stable abstention behavior, consistent citation tracking, and adaptive evidence retrieval. Evaluation follows a transparent and reproducible protocol inspired by PRISMA standards and Scopus-indexed benchmarking practices. S-AI Against Hallucinations provides a coherent, confidence-aware foundation for explainable and resource-efficient generative intelligence. It establishes a conceptual and operational bridge between statistical learning, symbolic reasoning, and biological homeostasis, paving the way for reliable and ethically governed AI systems.
This paper explores the integration of Green Artificial Intelligence (AI)—AI designed for energy efficiency and minimal environmental impact—with Islamic finance to advance sustainable resource management. Islamic finance, grounded in ethical principles such as justice, risk-sharing, and the prohibition of riba (interest), manages over $3.5 trillion in global assets as of 2024 [1]. However, its potential to address environmental sustainability remains underexplored. Green AI offers a solution by optimizing resource allocation in sectors critical to Muslim-majority economies, such as agriculture and renewable energy, while aligning with Maqasid al-Shariah (objectives of Islamic law), including hifz al-bi'ah (environmental preservation). Using a mixed-methods approach with case studies from the Middle East and Southeast Asia, we propose a novel framework that embeds Green AI into Sharia-compliant financial tools, demonstrating potential carbon emission reductions of up to 30% in optimized sukuk portfolios. This research contributes to theory by extending Maqasid al-Shariah to ecological stewardship, to practice by providing actionable AI models for Islamic banks, and to policy by recommending regulatory incentives for Green AI adoption. Our findings pave the way for mobilizing sustainable investments, bridging ethical finance with environmental sustainability.
Artificial intelligence (AI) is the powerful and the novelist tool used in a wide burden of applications and field worldwide, in respect to Saudi vision 2030, the use of AI in different disciplines become important and crucially adapted, computer science is the science of the AI as well. So, the different tools and methods utilized for enhancing education and training for these fields is so important to be implemented. This narrative review aims to give an overview about the tools, methods, and pathways for effective utilization of AI in training of computer sciences. In conclusion, and through intelligent tutoring, predictive analytics, and immersive simulations, learners can achieve deeper understanding, improved engagement, and enhanced employability. However, challenges such as data privacy, bias, infrastructure gaps, and resistance to adoption must be carefully addressed to ensure equitable and ethical utilization.
Cardiovascular disease (CVD) remains the leading cause of death globally, with roughly 17.9 million fatalities each year. Early and accurate diagnosis of heart disease is critical to improving patient outcomes. We propose OWE-CVD (Optimized Weighted Ensemble for Cardiovascular Disease), a new predictive framework that combines a weighted voting ensemble of three gradient boosting classifiers (XGBoost, LightGBM, and CatBoost) with explainable AI (XAI) techniques. We first addressed class imbalance using the Synthetic Minority Over-sampling Technique (SMOTE) and optimized each model’s hyperparameters using the Optuna framework with stratified 10-fold cross-validation. Ensemble weights were derived from cross-validated accuracy scores. On an independent test set, OWE-CVD achieved 94.44% accuracy with balanced precision, recall, and F1-scores across both classes. We applied SHAP and LIME to interpret the ensemble’s predictions at global and local levels. Overall, OWE-CVD demonstrates strong predictive performance while providing transparent decision support for heart disease diagnosis in clinical settings.
This paper presents a comprehensive critical review of contemporary technical solutions and approaches to artificial intelligence-based decision making systems in executive strategy scenarios. Drawing on systematic review of deployed technical solutions, algorithmic approaches, and empirical studies, this survey classifies and delineates the current decision support technology landscape and outlines future directions. Drawing on extensive review of current research and business application, the paper explains how AI technologies are redefining strategic decision frameworks in various industries. This survey contrasts machine learning algorithms, decision support architectures, and human-AI hybrid systems on various performance dimensions in a systematic way. The research points out prevailing trends such as the growth of augmented intelligence systems, the integration of predictive analytics with human intelligence, and new paradigms on ethics. Simulation results indicate that hybrid decision models that combine algorithmic precision with human intuition achieve 23% higher decision quality scores compared to algorithmic alone or human-alone approaches. The review outlines that effective executive strategy in the AI age calls for systematic organizational change involving technological infrastructure, leadership capability, and cultural adjustment.
The growing reliance on data-driven innovation in healthcare often collides with the critical need to protect patient privacy, creating a tension between progress and compliance. This study bridges that gap by introducing a Variational Autoencoder (VAE)-based framework to generate synthetic healthcare data that mirrors real-world datasets while ensuring privacy preservation. By leveraging synthetic EHRs created using the Synthea tool, the framework achieves a balance between statistical fidelity and data utility, enabling secure sharing and collaboration without compromising sensitive information. Through rigorous evaluation of distributional alignment and predictive performance, this work demonstrates the promise of synthetic data in unlocking the full potential of AI-driven healthcare solutions, offering a path to innovation that respects both privacy and progress.
One of the vital systems for the management of industrial infrastructure is SCADA (Supervisory Control and Data Acquisition). They are extensively applied in different industrial processes, particularly or energy, water, and transportation networks. These systems are principally efficient and unfailing when united with Artificial Intelligence (AI) technologies. The application of AI technologies in traditional SCADA systems creates many new opportunities. These technologies provide more accurate monitoring of processes, more effective control, increased security, and optimization of operations. But, due to their integration with modern Information Technologies and the Internet, these systems are more and more unprotected from cyber threats. Outdated security procedures are often unsatisfactory against these attacks. AI has emerged as a promising solution to enhance SCADA cybersecurity through anomaly detection, automated threat response, and predictive risk assessment. This article explores the applications of AI-driven cybersecurity in SCADA systems, highlighting the benefits and future research directions. Integrating artificial intelligence into SCADA security is crucial to ensuring resilience, reliability, and protection against both known and emerging cyber threats.
Five years before the release of ChatGPT, the world of Machine Translation (MT) was dominated by unimodal AI implementations, generally bilingual or multilingual AI models with only text modality. The era of Large Language Models (LLMs) led to various multimodal translation initiatives with text and image modalities, based on custom data engineering techniques that introduced expectations for improvement in the field of MT when using multimodal options. In our work, we introduced a first of its kind AI multimodal translation with four modalities (text, image, audio and video), from English towards a low resource language and vice-versa. Our results confirmed that multimodal translation generalizes better, always brings improvement to unimodal text translation, and superior performance as the number of unseen samples increases. Moreover, this initiative is a hope for worldwide low resource languages for which the use of non-text modalities is a great solution to data scarcity in the field.
This paper presents an enhanced framework to strengthening privacy and security in Retrieval-Augmented Generation (RAG)-based AI applications. With AI systems increasingly leveraging external knowledge sources, they become vulnerable to data privacy risks, adversarial manipulations, and evolving regulatory frameworks. This research introduces cutting-edge security techniques such as privacy-aware retrieval mechanisms, decentralized access controls, and real-time model auditing to mitigate these challenges. We propose an adaptive security framework that dynamically adjusts protections based on contextual risk assessments while ensuring compliance with GDPR, HIPAA, and emerging AI regulations. Our results suggest that combining privacy-preserving AI with governance automation significantly strengthens AI security without performance trade-offs.
This paper examines the potential of Large Language Models (LLMs) in revolutionizing lead qualification processes within sales and marketing. We critically analyze the limitations of traditional methods, such as dynamic branching and decision trees, during the lead qualification phase. To address these challenges, we propose a novel approach leveraging LLMs. Two methodologies are presented: a single-phase approach using one comprehensive prompt and a multi-phase approach employing discrete prompts for different stages of lead qualification. The paper highlights the advantages, limitations, and potential business implementation of these LLM-driven approaches, along with ethical considerations, demonstrating their flexibility, maintenance requirements, and accuracy in lead qualification.
Generative AI and the potential it carries has enabled businesses to incorporate intelligent automation and optimization across domains. With this paper we present a methodology to enhance b2b pricing and deal negotiation process where langchain framework and agents could be used to summarize critical information regarding businesses (B2B customers)in real time to indirectly calculate their willingness to pay in the form of a score generated by reasoning LLMs. This score will acts as an indicator to suggest the customer’s ability to accept the price point being negotiated in the given period of time based on their financial health, market sentiments and internal to company performance.
Artificial Intelligence (AI) systems are transforming various industries, offering new opportunities and efficiencies. However, alongside these benefits, the development and deployment of AI raise significant ethical considerations. This paper examines the ethical issues surrounding AI, including bias, transparency, privacy, accountability, and societal impact. It proposes guidelines for ensuring the responsible use of AI technologies, emphasizing the importance of prioritizing ethical principles such as fairness, transparency, and accountability. However, the widespread adoption of AI also brings forth a range of ethical challenges that must be addressed to ensure that these technologies are developed and deployed responsibly. One of the most pressing ethical concerns is the issue of bias in AI systems. AI algorithms are often trained on large datasets that may contain historical biases, which can lead to discriminatory outcomes when these systems are used in real-world applications.
This paper introduces the Boswell Test, a new benchmark for artificial intelligence (AI) that builds upon the legacy of the Turing Test. Inspired by James Boswell's insight into Samuel Johnson, it evaluates AI's potential to evolve from mere assistants into indispensable companions with human-like understanding. The test is divided into Test-A (mastery of human nuances) and Test-B (critical thinking). This study presents an initial implementation of Test-B, focusing on AI chatbots' analysis of global AI policies and calculates a Boswell Quotient using metrics of normalized median grades, accuracy, consistency, userfriendliness, and truthfulness to reveal strengths and limitations of current AI, paving the way for more humanistic advanced systems.
With the increasing amount of data available, recommendation systems are important for helping users find relevant content. This paper introduces a movie recommendation system that uses user profiles and machine learning techniques to improve the user experience by offering personalized suggestions. We tested different machine learning methods, including k nearest neighbors (KNN), support vector machines (SVM), and neural networks. We used several datasets, such as MovieLens and Netflix Prize, to check how accurate the recommendations were and how satisfied users were with them.
Accurately identifying at-risk students in higher education is crucial for timely interventions. This study presents an AI-based solution for predicting student performance using machine learning classifiers. A dataset of 208 student records from the past two years was preprocessed, and key predictors such as midterm grades, previous semester GPA, and cumulative GPA were selected using information gain evaluation. Multiple classifiers, including Support Vector Machine (SVM), Decision Tree, Naive Bayes, Artificial Neural Networks (ANN), and k-Nearest Neighbors (k-NN), were evaluated through 10-fold crossvalidation. SVM demonstrated the highest performance with an accuracy of 85.1% and an F2 score of 94.0%, effectively identifying students scoring below 65% (GPA < 2.0). The model was implemented in a desktop application for educators, providing both class-level and individual-level predictions. This userfriendly tool enables instructors to monitor performance, predict outcomes, and implement timely interventions to support struggling students. The study highlights the effectiveness of machine learning in enhancing academic performance monitoring and offers a scalable approach for AI-driven educational tools.
While Vision Transformers (ViTs) have revolutionized computer vision with their exceptional results, they struggle to balance processing speed with visual detail preservation. This tension becomes particularly evident when implementing larger patch sizes. Although larger patches reduce computational costs, they lead to significant information loss during the tokenization process. We present AE-ViT, a novel architecture that leverages an ensemble of autoencoders to address this issue by introducing specialized latent tokens that integrate seamlessly with standard patch tokens, enabling ViTs to capture both global and fine-grained features. Our experiments on CIFAR-100 show that AE-ViT achieves a 23.67% relative accuracy improvement over the baseline ViT when using 16×16 patches, effectively recovering fine-grained details typically lost with larger patches. Notably, AE-ViT maintains relevant performance (60.64%) even at 32×32 patches. We further validate our method on CIFAR-10, confirming consistent benefits and adaptability across different datasets. Ablation studies on ensemble size and integration strategy underscore the robustness of AE-ViT, while computational analysis shows that its efficiency scales favorably with increasing patch size. Overall, these findings suggest that AE-ViT provides a practical solution to the patch-size dilemma in ViTs by striking a balance between accuracy and computational cost, all within a simple, end-to-end trainable design.
The effectiveness of ensemble learning in improving prediction accuracy and estimating uncertainty is wellestablished. However, conventional ensemble methods often grapple with high computational demands and redundant parameters due to independent network training. This study introduces the Divergent Ensemble Network (DEN), a novel framework designed to optimize computational efficiency while maintaining prediction diversity. DEN achieves superior predictive reliability with reduced parameter overhead by leveraging shared representation learning and independent branching. Our results demonstrate the efficacy of DEN in balancing accuracy, uncertainty estimation, and scalability, making it a robust choice for realworld applications.
Artificial intelligence is transforming various fields, including accounting, by representing a significant technological innovation. Artificial intelligence combines hardware and software to simulate human cognitive processes, enabling machines to perform complex tasks such as learning, reasoning, and decision-making. This paper explores the advantages and disadvantages of integrating artificial intelligence into accounting practices. While artificial intelligence presents numerous benefits for accountants, it also introduces challenges that must be addressed. The paper also contributes to the expanding knowledge base on artificial intelligence in accounting by offering practical recommendations for accountants on effectively adopting artificial intelligence. Even with the challenges presented from integrating artificial intelligence in accounting, such integration offers considerable efficiency gains. This positions artificial intelligence as a strategic investment for organizations aiming to improve the performance and effectiveness of their accounting departments.
Smart contracts, integral to blockchain ecosystems, enable decentralized applications to execute predefined operations without intermediaries. Their ability to enforce trustless interactions has made them a core component of platforms such as Ethereum. Vulnerabilities such as numerical overflows, reentrancy attacks, and improper access permissions have led to the loss of millions of dollars throughout the blockchain and smart contract sector. Traditional smart contract auditing techniques such as manual code reviews and formal verification face limitations in scalability, automation, and adaptability to evolving development patterns. As a result, AI-based solutions have emerged as a promising alternative, offering the ability to learn complex patterns, detect subtle flaws, and provide scalable security assurances. This paper examines novel AI-driven techniques for vulnerability detection in smart contracts, focusing on machine learning, deep learning, graph neural networks, and transformer-based models. This paper analyzes how each technique represents code, processes semantic information, and responds to real world vulnerability classes. We also compare their strengths and weaknesses in terms of accuracy, interpretability, computational overhead, and real time applicability. Lastly, it highlights open challenges and future opportunities for advancing this domain.
