Generative Artificial Intelligence (Generative AI) has rapidly transitioned from an experimental technology into a working companion for researchers across disciplines. Tools built on large language models (LLMs) and related generative architectures can now draft text, summarise literature, suggest hypotheses, generate code, and assist with data interpretation, changing how research is planned, conducted, and communicated. This paper examines how Generative AI is being applied to enrich the research process, drawing on recent academic literature published between 2023 and 2025.

Finding a free parking space in busy cities, shopping malls, offices, and campuses wastes time and fuel, adds to traffic inside parking areas, and frustrates drivers. Traditional parking management relies on human attendants, tickets, or a sensor fitted at every slot, which is slow, costly, and hard to maintain. Computer vision offers a practical alternative: the CCTV cameras that already watch a parking area can be combined with artificial intelligence to find vehicles and decide, in real time, which slots are occupied and which are vacant.

Prompt engineering has emerged as the primary interface through which practitioners steer the behaviour of large language models (LLMs) without updating any model weights. Since the introduction of few-shot in-context learning, a wide family of prompting strategies has been proposed, ranging from simple zero-shot instructions to elaborate reasoning frameworks and agentic, tool-using prompts.

Websites that can be used by language-model agents need more than a collection of API endpoints. They need compact observations, unambiguous actions, current state, authorization boundaries, accessible human fallbacks, and evidence that can be checked after execution. This review examines ten supplied studies covering agentic AI, Model Context Protocol (MCP), structured web access, human–AI dialogue, tool security, and domain-specific orchestration.

Artificial intelligence (AI) and deep learning (DL) are progressively used in social network analysis (SNA) to improve topology classification, link prediction, and community detection, thereby reducing manual analytical errors. This study addresses the use of graph neural networks, especially the Stacked Graph Convolutional Network (Stacked GCN-GAT) model, in social structural diagnosis and node interaction classification.

LoRa devices send payloads of a few bytes, so every authentication-tag byte is a large share of each frame. NIST SP 800-232 standardises Ascon-AEAD128 for constrained devices, permits tags shorter than 64 bits only after a careful risk analysis, and bounds the verification failures tolerated per key. This study performs that analysis for LoRa uplinks. An analytical model links the tag length to time on air, forgery probability against a channel-time-limited attacker, the attacker airtime needed to exhaust the failure budget and force a key update, benign budget consumption by corrupted frames that pass the 16-bit LoRa CRC, and packet errors.