The Impact of ChatGPT and DeepSeek on Academic Literature: A Multi-Disciplinary Analysis

Abstract

This paper examines the impact of AI-driven large language models, such as ChatGPT and DeepSeek, on academic publishing in three fields: cybersecurity, optimization, and robotics. A comparative review has been performed between 2019 and 2025, dividing the period into the pre-AI era (2019–2021) and the post-AI deployment era (2022–2025). Investigating trends in publications, writing quality, citation behavior and linguistic characteristics, a revolutionary increase was observed in publications for AI-enabled healthcare, cybersecurity, and robotics, where the latter has increased by 585%. According to Flesch-Kincaid scores, which measure how easy or difficult a text is to read based on sentence length and word complexity, the papers created have shown more linguistic maturity. In addition, they become more challenging to read, but each type of document in each research field expresses specific profiles of change. For example, cybersecurity writings tend toward a formal tone; optimization papers turn toward being more demanding of precision and conciseness, while robotics research employs AI-specific terminology and increasingly uses AI-influenced subject matter. AI-based tools, such as ChatGPT and DeepSeek, can optimize efficiency, accessibility, and linguistic prowess. However, their traps suggest an increase in fake citations and a more sophisticated overreliance on machine-generated content. The work proposes some pragmatic recommendations, thus providing a path to reconcile innovation and integrity.

Ähnliche Arbeiten