Stevan STANKOVSKI
Generative artificial intelligence (GenAI) is transforming modern higher education by providing new opportunities for teaching, learning, and research. In the context of mechatronics and industrial engineering education, GenAI tool applications in programming and industrial automation instruction are particularly interesting. These fields require developing complex cognitive skills, such as abstract thinking, logical reasoning, and understanding physical processes and design conventions. The question is how GenAI tools affect the development of these competencies. Do they speed up and deepen the development of these skills, or do they replace cognitive effort without fostering genuine understanding? This paper analyzes the use of GenAI tools in an academic setting, focusing on two technical areas: generating program code and designing electropneumatic schemes.
Duško ČUČKOVIĆ, Dejan IGOV
The aim of this paper is to analyze theoretical perspectives on the relationship between TQM and organizational innovation performance based on a review of the relevant literature. The analysis indicates that the relationship between TQM and organizational innovation performance is neither universally synergistic nor inherently conflictual, but rather depends on the dimensions of TQM, the type of innovation, and the organizational context. Particular emphasis is placed on distinguishing between soft and hard TQM practices, as well as between incremental and radical innovation. It is concluded that TQM can provide an important organizational foundation for innovation, while the effects of individual TQM practices depend on how they are implemented and on the specific requirements of different types of innovation.
Tanja KRUNIĆ
The main focus of this research is the comparison of human-written, AI-generated, and humanized texts optimized for search engines. For two distinct topics, appropriate keyword lists were created, and based on them, human-authored texts were written and compared to texts generated by the three most popular language model services: ChatGPT, Google Gemini, and Microsoft Copilot. The study tests the originality value of AI-generated text produced by various generators, as well as human-written content. The results indicate that both groups of texts are evaluated as original by two distinct plagiarism checkers. The research analyzes the ability of AI-generated content to bypass detection in its original version and after humanization. The findings indicate that content generated by Google Gemini bypasses AI-content detectors more easily than that generated by ChatGPT and Microsoft Copilot, especially after human-ization. On the other hand, ChatGPT has shown the lowest probability of bypassing AI-content detectors. This research also analyzes how humanization affects text readability. The results indicate that humanization can further decrease readability, producing text that is less engaging for readers. Additionally, a hybrid quantitative and qualitative analysis was performed. The quantitative metrics map the bounda-ries of detection score shifts, while the qualitative evaluation isolates specific linguistic anomalies, identifying repetitive patterns like structural freezing and syntactic inversion. Ultimately, the study concludes that no single AI-text gener-ator, AI-content detector, or humanization tool is perfect. Human oversight remains very important in obtaining at-tractive, readable web content.