20.500.12556/DKUM-98583 Primerjalna analiza orodij umetne inteligence za generiranje programske kode: učinkovitost, varnost in vpliv na razvoj aplikacij učinkovitost, varnost in vpliv na razvoj aplikacij Comparative analysis of artificial intelligence tools for code generation: effectiveness, security, and impact on application development diplomsko delo Diplomsko delo obravnava uporabo generativne umetne inteligence pri generiranju programske kode. Namen raziskave je bil primerjati orodja ChatGPT, GitHub Copilot, Microsoft Copilot, DeepSeek in Grok pri reševanju manjših implementacijskih nalog v programskem jeziku Python. Orodja so bila primerjana glede na natančnost, robustnost, berljivost in čas generiranja začetne programske rešitve. Raziskava je vključevala eksperimentalno testiranje petih programskih nalog in anketni ter intervjujski del. Rezultati kažejo, da se orodja razlikujejo predvsem pri obravnavi napak, strukturiranosti kode in hitrosti generiranja. Orodji DeepSeek in GitHub Copilot sta ustvarili bolj robustne in strukturirane rešitve, orodje Grok pa je izstopalo po krajšem času generiranja. Rezultati ne predstavljajo neposredne mere celotne produktivnosti razvijalca, saj celoten čas dokončanja naloge ni bil merjen. The thesis examines the use of generative artificial intelligence in software code generation. The aim of the research was to compare the tools ChatGPT, GitHub Copilot, Microsoft Copilot, DeepSeek, and Grok in solving smaller implementation tasks in the Python programming language. The tools were compared in terms of accuracy, robustness, readability, and the time required to generate the initial software solution. The research included experimental testing of five programming tasks, as well as a survey and interviews. The results show that the tools differ mainly in error handling, code structure, and generation speed. DeepSeek and GitHub Copilot generated more robust and structured solutions, while Grok stood out with a shorter generation time. The results do not represent a direct measure of overall developer productivity, as the total time required to complete the task was not measured. umetna inteligenca generiranje kode produktivnost razvijalcev varnost kode artificial intelligence code generation developer productivity code security true true false [D. Novak] Slovenski jezik Angleški jezik Diplomsko delo/naloga 2026-06-22 17:18:41 2026-08-18 09:19:38 2026-09-01 03:08:07 0000-00-00 00:00:00 2026 0 Maribor 0 Maribor 1 spletni vir (1 datoteka PDF (IX, 75 str.)) 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.8(043.2) 289390083 VS_Novak_David_2026.pdf VS_Novak_David_2026.pdf 1 5A5F498A68FA580BEA5078E448F5301A 4b7b2457dbc3566a7de54ab9fb03b86b76346c88fb53c3b54529165da60b07d0 a6615082-7393-11f1-a1c0-00155d580e34 https://dk.um.si/Dokument.php?lang=slv&id=207676 VS_Novak_David_2026.zip VS_Novak_David_2026.zip 1 ED5078416F76B132EA2C53264A8393F0 a9ffa8995d4307f0c0bb820f534fab15c8b46a1d6910cca3e0f46bdbb6a9fb9a bca7ae36-7393-11f1-a1c0-00155d580e34 https://dk.um.si/Dokument.php?lang=slv&id=207678 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0