Special Issue on Computing, Engineering and Multidisciplinary Sciences 2025

This article enhances discussions on Generative Artificial Intelligence (GenAI) and prompt engineering by exploring critical pitfalls and industry-specific advantages. It begins with a foundational overview of AI evolution, emphasizing how generative models such as GANs, VAEs, and Transformers have revolutionized language processing, image generation, and drug discovery. Prompt engineering is highlighted as a key methodology for directing model outputs with precision and ethical awareness, enabling applications in Natural Language Processing (NLP), content personalization, and decision support. The revised sections detail how prompt engineering can be misapplied, underscoring common errors like overly restrictive or ambiguous prompts that compromise GenAI’s accuracy, ethicality, and creative capacity. Equally, the paper showcases high-impact use cases in finance, education, healthcare, and beyond, illustrating how carefully formulated prompts can strengthen risk detection, enhance student learning, improve clinical decision-making, and foster product innovation. The expanded discussion of industry alignment illustrates the tangible value these techniques offer across diverse sectors, ultimately reinforcing the notion that prompt engineering is central to maximizing GenAI’s transformative potential. Future directions address emerging trends, from multimodal fusion and domain-specific fine-tuning to adaptive prompt designs that leverage real-time user feedback, further solidifying the role of responsible prompt engineering in shaping the next generation of intelligent and ethically aligned AI solutions.

Providing timely and meaningful feedback in photography education is challenging, particularly in large classes where manual assessment can delay skill development. This paper presents M-Stock, an AI-based automated photo evaluation system that uses Convolutional Neural Networks (CNNs) to assess student photography assignments on web browser. M-Stock evaluates both technical aspects (such as lighting, composition, and exposure) and creative elements, providing students with real-time, formative feedback. The system was trained on a diverse dataset, including student submissions and commercial standards, achieving an overall accuracy of 97.18% with an average prediction speed of 46.1 milliseconds per image. Experiments assessed the system’s performance across varying resolutions and batch sizes, confirming its scalability and suitability for real-time classroom use. Additionally, a pilot study with students indicated that M-Stock’s feedback positively impacted their technical skills and encouraged self-directed learning. The results demonstrate M-Stock’s potential as a transformative tool for photography education, combining high accuracy, immediate feedback, and pedagogical value to support continuous learning. Future improvements will focus on refining creative assessments and expanding the system’s applicability to other visual arts disciplines.

Artificial Intelligence (AI) is reshaping the global shipbuilding sector, yet existing maturity models fail to capture the domain-specific complexities of this capital-intensive industry. This study reviews over 50 AI maturity models and introduces a specialized framework tailored for shipbuilding. The proposed model outlines four progressive stages—Beginner, Innovation, Integration, and Expert—across eight key dimensions: culture, resilience, sustainability, strategy, customer focus, organizational integration, connectivity, and production efficiency. A hybrid benchmarking approach involving comparative analysis of major shipbuilders such as China State Shipbuilding Corporation(CSSC), General Dynamics National Steel and Shipbuilding Company(NASSCO), and Hyundai Heavy Industries(HHI), as well as synthesis from literature, was used to validate the relevance and coverage of each dimension. The framework provides a roadmap for operational modernization and links digital maturity to measurable outcomes such as delivery timelines, production scalability, and environmental performance. Policy recommendations highlight the need for targeted investments, workforce reskilling, and public-private collaboration to enable sustainable and AI-enabled growth in the U.S. shipbuilding sector.

This study presents a novel alternative to traditional Net Energy Metering (NEM) by proposing a set of innovative pricing schemes for solar customers participating in utility-led grid service programs through the aggregation of Distributed Energy Resources (DERs). Grounded in cooperative game theory, the proposed framework facilitates equitable and efficient value allocation among key stakeholders, namely customers, utilities, and aggregators—based on their respective marginal contributions to grid performance and system cost reductions. In contrast to legacy NEM structures, which typically remunerate customers at retail rates and inadequately incentivize storage adoption, load flexibility, or temporal optimization, this approach enables new revenue opportunities by embedding DERs within coordinated grid service portfolios. The pricing mechanisms developed herein are centered on two critical grid services: energy arbitrage and peak load management. These services are provisioned by the excess capacity of customer-owned DERs, particularly rooftop photovoltaic systems and behind-the-meter battery storage. Through the implementation of a Grid Services Set (GSS) and a complementary Grid Services Rider (GSR) tariff structure, participating customers voluntarily permit automated utility coordination of their devices in return for performance-based compensation. An integrated optimization algorithm co-optimizes DER dispatch across both distribution-level operational requirements and real-time wholesale market opportunities, such as those found in the Energy Imbalance Market. This enables strategic charging during periods of surplus or negative pricing and discharging during price peaks. The proposed model contributes to the advancement of Non-Wires Alternatives (NWAs) by reducing reliance on conventional infrastructure upgrades and enhancing grid flexibility and resilience. It also offers a regulatory-aligned pathway for harmonizing DER integration with utility planning objectives, renewable energy targets, and climate adaptation strategies. By fostering a cooperative paradigm between utilities and customers, the framework promotes prosocial grid behavior, scalable DER participation, and innovation in the evolving landscape of decentralized energy systems.

Persistent anomalies in modern photovoltaic (PV) systems present a formidable challenge, impeding optimal power output and system resilience. Artificial Intelligence (AI) has surfaced as a game-changing solution, yet existing research has merely scratched the surface of solar panel prognosis, leaving a critical void in leveraging AI’s explainable nature and active learning capabilities. This pioneering study investigates AI methods for detecting and classifying critical faults in PV systems, pushing the boundaries of innovative methodologies for fault identification. We acknowledge that the opacity of AI methods can hinder their adoption, particularly among practitioners, thus emphasizing Explainable AI (XAI) in an exhaustive bibliometric analysis. This study showcases authors who thoroughly detail their development processes and underscores the indispensable role of human/expert interaction in active learning for labeling the most informative data. Our findings unveil a glaring underutilization of XAI in the solar panel domain, with China at the forefront of this field. This leadership is likely attributed to the robust research focus in Chinese universities and China’s position as the world’s leading solar panel producer. We delve into the potential role of human/expert involvement in designing and deploying deep learning predictive applications, highlighting methods that harmoniously integrate practical knowledge from human end-users through active learning. Our methodology encompasses extensive data collection, bibliometric analysis of collaborations between entities, researchers, and nations, and an examination of the most prevalent persistent faults. We conclude by strongly advocating for future studies to address the underutilization of XAI and active learning in AI-based defect prediction. Bridging this gap is crucial for pinpointing the root causes of solar panel defects and enhancing prognosis, positioning this research as indispensable for both scientists and industry professionals at the forefront of PV technology.

Ensuring the quality of higher education in Brazil presents a complex challenge, intensified by factors that directly affect students’ academic performance. The pervasive influence of social media and the overconsumption of superficial digital content undermine students’ ability to engage in deep comprehension, critical thinking, and the practical application of knowledge. Furthermore, inadequate preparation during the preceding educational years hinders students’ ability to adapt to the academic demands of higher education, leading to difficulties in academic progression and increased dropout rates. In view of the above, this paper explores the potential of Machine Learning models (ML) in predicting the academic performance of higher education students within the Ânima Educação ecosystem, Brazil. The contribution of this work is the development of an artificial intelligence-based assessment tool called AILA that recommends personalized study content for fundamental skills such as Portuguese and Mathematics, based on the psychometric profile of each student. This approach aims to optimize the learning process by addressing individual needs, enhancing academic performance, and overcoming the challenges faced by students in the contemporary educational landscape. Psychometric profile data were collected from approximately 41,296 incoming students of the Ânima Educação ecosystem universities on the following dimensions: learning, social intelligence, emotional management, socio-emotional skills, teaching method, and knowledge area of the students. The AILA ML models presented good results in predicting students’ basic skills performance in the binary and regression approaches. Specifically, the CatBoost model showed an accuracy of 0.74 in predicting scores on the Portuguese and Mathematics and Logical Reasoning proficiency tests.