Future Digital Technologies and Artificial Intelligence

Digital dependency and its consequences for human well-being

2025-04-11T05:43:21+00:00

This article examines how digital technology affects the thinking and emotional well-being of young people. Spending too much time on screens can make it harder to concentrate, remember important information, get enough sleep, and manage emotions. Research shows that frequent use of smartphones and social media leads to serious problems known as digital dementia and brain rot. These habits can change brain function, cause unhealthy behaviors, and mental disorders. The negative effects can be reduced by setting limits on device use and encouraging controlled usage. Simple steps such as putting phones away during meals, reading printed books, and spending more time outdoors have been shown to have positive impacts on health. Taking regular breaks from digital content also helps the mind stay clear. The complex problem of the digital era can only be solved with collective action. All interested parties should cooperate on the policies and guidelines for using digital tools to protect mental health in the long run.

A study on adversarial attacks in Deep Learning-based traffic signal recognition for autonomous vehicles

2025-07-05T17:35:53+00:00

Autonomous vehicles are gradually occupying the streets and are expected to become ubiquitous in the near future. However, recent incidents involving these vehicles have raised serious concerns about their safety, particularly regarding the reliability of their onboard machine learning systems. In this paper, we expose a critical yet underexplored vulnerability—misclassifying street signs as traffic lights—by conducting a targeted white-box adversarial attack. To the best of our knowledge, this specific vulnerability has not been addressed in the existing literature. We craft adversarial examples using the Fast Gradient Sign Method (FGSM) to generate minimal perturbations that can deceive a state-of-the-art image classification model, Inception-V3, trained on the ImageNet dataset. We also introduce a custom dataset consisting of real-world street sign and traffic light images to test the attack under more domain-specific conditions. Our evaluation metrics include attack success rate, Structural Similarity Index (SSIM), and L2 distance, with our method achieving a 100% success rate in misclassification. These results highlight the pressing need to design robust defenses against adversarial attacks in safety-critical systems. We further discuss technical challenges, potential defenses such as adversarial training and obfuscated gradients, and directions for future research to enhance the resilience of deep learning systems in autonomous vehicles.

Constructing enterprise talent heterogeneous information networks for key talent identification

2025-10-01T16:34:43+00:00

In organizational networks, where employee performance is dependent on strategic positioning and collaborative relationships across diverse workplace ecosystems, traditional enterprise talent identification systems fall short in capturing complex multi-relational dynamics. In order to accurately identify key talent through meta-path guided feature extraction and attention-based embedding mechanisms, this research suggests a Heterogeneous Information Network (HIN) framework that uses Graph Neural Networks (GNNs) to model employees, projects, departments, and skills as interconnected entities. The approach uses Heterogeneous Graph Attention Networks (HAN) for talent assessment and combines attribute-driven performance indicators, structural centrality measures, and semantic relationship patterns into a single learning framework. Compared to traditional Human Resource (HR) methods, which scored 72% precision and 68% recall, the experimental evaluation, which used enterprise data with 2,847 employees across 156 departments, shows improvements over current approaches, achieving 91% precision and 89% recall with a Normalized Discounted Cumulative Gain (NDCG) of 0.834. With domain expert validation confirming 87% agreement between algorithmic recommendations and professional assessments, the framework identifies high-potential employees who exhibit knowledge brokerage roles and cross-functional collaboration capabilities that traditional performance metrics overlook. With implications for strategic human capital optimization, these contributions position HINs as a paradigm shift for enterprise talent management.

Empowering vocational education in Africa through AI and deep learning technologies

2025-10-02T04:36:58+00:00

Vocational schools in Sub-Saharan Africa face critical challenges, including inadequate equipment, insufficient funding, and curricula misaligned with industry needs. This study explores how artificial intelligence (AI) and deep learning address these challenges through empirical research in Nigeria and Kenya. The research tests adaptive learning systems with 742 students, comparing AI-enhanced with traditional methods. Results demonstrate 68% faster skill acquisition (t=4.82, p<0.01, d=0.68) and improved job readiness (χ²=18.3, p<0.05). Model compression to 45-75MB enables deployment on basic smartphones while maintaining 92% accuracy. Implementation includes mobile-first platforms tested in three Nigerian vocational centers and automated skill recognition systems deployed in two Kenyan technical schools. The findings confirm that properly localized AI solutions can transform vocational training in resource-limited contexts, though sustainability challenges remain.