Conceptual And Applied Aspects of Artificial Intelligence – An Analysis of Ai Capabilities, Limitations, And Prospects in Modern Technologies

Аннотация

This article presents a comprehensive analysis CITATION of artificial intelligence and its impact on various aspects Sergiu Metgher. (2025). Conceptual And Applied Aspects of Artificial Intelligence – An Analysis of Ai Capabilities, Limitations, And Prospects in of sustainable development. AI is actively utilized to Modern Technologies. The American Journal of Engineering and enhance decision-making, automate processes, and Technology, 7(8), 21–29. https://doi.org/10.37547/tajet/Volume07Issue08-03 optimize numerous fields of activity. However, its COPYRIGHT integration into critical domains such as sustainable © 2025 Original content from this work may be used under the terms development, public administration, and cultural of the creative commons attributes 4.0 License. innovation raises concerns regarding accessibility, inclusivity, and resource redistribution. The study examines key aspects of artificial intelligence, its classification—including narrow and general AI—and the technologies employed, such as machine learning, natural language processing, and computer vision. Special attention is given to AI’s relationship with sustainable development goals and its role in advancing innovative solutions in social and environmental spheres. The article also explores the ethical, social, and cultural consequences of AI implementation, emphasizing the necessity of developing responsible, transparent, and sustainable systems that align with international standards and ensure long-term societal well-being. This research may be of interest to a broad audience of specialists and researchers engaged in fields related to AI development, its applications, and its influence on societal processes. Keywords: artificial intelligence, sustainable development, machine learning, ethics, innovation, social development, ecology, sustainable development goals. The American Journal of Engineering and Technology 21 https://www.theamericanjournals.com/index.php/tajet Introduction Materials and Methods As artificial intelligence moves beyond its experimental The scientific disciplines underlying artificial intelligence capabilities and becomes an integral part of modern are diverse and encompass machine learning (ML), civilization’s daily functions, the implications of its natural language processing (NLP), computer vision, deployment become increasingly complex to predict. expert systems, robotics, and automated reasoning. The Governments, businesses, and research institutions interconnectivity of these fields underscores AI’s broad leverage AI-driven systems to enhance decision-making, applicability across various sectors. optimize logistics networks, and even address Notably, machine learning serves as the cornerstone of macroeconomic challenges. Economic projections modern AI advancements, implementing the suggest that by 2030, AI’s global economic impact will fundamental concept of artificial intelligence through reach approximately $15 trillion, underscoring its sophisticated algorithms that iteratively enhance potential as a catalyst for systemic transformations [1]. predictive accuracy and decision-making efficiency. This At the same time, AI integration into areas such as domain includes applications such as speech sustainable development, public governance, and recognition, affective computing, economic forecasting, cultural innovation raises significant concerns regarding and industrial process optimization, exemplifying the accessibility, inclusivity, and the redistribution of wealth universality and adaptability of ML paradigms across a and labor in an increasingly automated world. wide range of practical implementations. A distinctive Artificial intelligence, broadly defined as "the capability feature of machine learning is its methodological of a digital computer or a computer-controlled robot to stratification, divided into supervised learning (SL), perform tasks commonly associated with intelligent unsupervised learning (UL), semi-supervised learning beings," extends beyond simple automation. Instead, it (SSL), reinforcement learning (RL), and deep learning represents a complex interplay of software and (DL) [1]. Supervised learning, based on labeled datasets, hardware methodologies designed to replicate—or at facilitates classification and regression tasks, whereas least approximate—human cognition and behavior. unsupervised learning uncovers hidden structures and Fundamentally, AI is divided into narrow (weak) AI and anomalies within unlabeled data distributions. general (strong) AI. Practical applications predominantly Reinforcement learning optimizes decision-making fall into the former category, comprising specialized, policies through iterative interactions with dynamic highly optimized systems that demonstrate environments, while deep learning, leveraging performance levels sufficient for real-world multilayered neural architectures, excels in feature implementation but lack true autonomous reasoning. abstr