20.500.12556/RUL-184521 Generativna umetna inteligenca kot podpora personaliziranemu učenju kemije: empirična analiza učnih učinkov, interesa in stališč dijakov. magistrsko delo Generative artificial intelligence as support for personalized chemistry learning: an empirical analysis of learning outcomes, interest, and student attitudes Kemija po izobraževalni vertikali pogosto velja za zahtevno in abstraktno, zato učitelji pouk kemije med drugim podpirajo z IKT učnimi pristopi. Ob pojavu generativne umetne inteligence (GenUI) se odpira novo vprašanje: ali lahko dotična orodja smiselno in odgovorno obogatijo tako učenje kot tudi poučevanje kemije v srednji šoli. V magistrskem delu je bila empirično preverjena primernost uporabe GenUI kot podpore dijaku v učnem procesu. Raziskano je bilo, ali se priprava na laboratorijsko delo z uporabo GenUI v obravnavanem učnem kontekstu povezuje z boljšim razumevanjem kemijskih učnih vsebin, višjimi učnimi dosežki in večjo trajnostjo znanja ter kako se povezuje z interesom, zaznano kompetentnostjo dijakov in njihovimi stališči do uporabe GenUI pri pouku kemije. V raziskavo je bil vključen namenski vzorec dijakov 2. letnika gimnazije v osrednjeslovenski regiji (N = 27), ki so pri pouku kemije izvedli dve laboratorijski vaji iz tematskega učnega sklopa Ravnotežja v vodnih raztopinah. Dijaki so bili razdeljeni v dve heterogeni skupini. Prva skupina se je na prvo laboratorijsko vajo pripravljala na tradicionalen način, druga pa tradicionalno z dodano personalizirano podporo GenUI. Pri drugi laboratorijski vaji sta se načina priprave skupin zamenjala. Po vsaki vaji so dijaki reševali preizkus znanja in izpolnjevali vprašalnike s trditvami Likertovega tipa za merjenje situacijskega interesa, zaznane kompetentnosti in stališč do uporabe GenUI. Po dveh tednih je bil izveden odloženi preizkus znanja, izbranih deset dijakov pa je sodelovalo v polstrukturiranih intervjujih o izkušnjah z uporabo GenUI pri pouku in učenju kemije. Rezultati so pokazali, da so dijaki pri pripravi z uporabo GenUI na obeh neposrednih preizkusih znanja dosegli statistično značilno boljše rezultate kot pri tradicionalni pripravi. Podoben vzorec se je pokazal tudi pri odloženih preizkusih znanja, kar nakazuje na možnost, da je bila priprava z uporabo GenUI v obravnavanem kontekstu povezana z boljšim ohranjanjem znanja. Dijaki so ob uporabi GenUI poročali o višjem situacijskem interesu in višji zaznani kompetentnosti pri laboratorijskem delu. Pri stališčih do GenUI so bile pozitivne spremembe izrazite predvsem pri zaznavanju uporabnosti GenUI za učenje in pouk kemije. Kvalitativni rezultati so pokazali, da so dijaki GenUI večinoma doživljali kot hitro, dostopno in uporabno učno podporo, ki jim je pomagala pri razumevanju pojmov, preverjanju odgovorov, razlagi po korakih in pripravi na laboratorijsko delo. Obenem so opozorili na omejitve, kot so možnost napačnih ali premalo osredotočenih odgovorov, potreba po ustreznem predznanju, zahtevnost oblikovanja kakovostnih vprašanj ter dileme glede meje med dovoljeno učno podporo in nedovoljeno bližnjico. Chemistry is often perceived as a demanding and abstract subject throughout the educational pathway. For this reason, chemistry teachers, among other strategies, support instruction through ICT-based teaching approaches. With the emergence of generative artificial intelligence (GenAI), a new question arises: whether such tools can meaningfully and responsibly enrich both the learning and teaching of chemistry in upper-secondary education. This master’s thesis empirically examined the suitability of using GenAI as a form of student support in the learning process. The study investigated whether GenAI contributed to a better understanding of chemistry learning content and improved learning outcomes, whether it enhanced the durability of knowledge, how it influenced students’ interest and perceived competence, and what attitudes students held toward GenAI and its use in chemistry lessons. The study included a purposive sample of second-year general upper-secondary school students from the Central Slovenia region (N = 27). As part of chemistry lessons, the students carried out two laboratory exercises within the thematic unit Equilibria in Aqueous Solutions. They were divided into two heterogeneous groups. For the first laboratory exercise, one group prepared in the traditional way, while the other prepared traditionally with the addition of personalised GenAI support. For the second laboratory exercise, the preparation methods were reversed between the groups. After each laboratory exercise, the students completed a knowledge test and questionnaires consisting of Likert-type statements measuring situational interest, perceived competence, and attitudes toward the use of GenAI. Two weeks later, a delayed knowledge test was administered, and ten selected students participated in semi structured interviews about their experiences with using GenAI in chemistry learning and instruction. The results showed that, when preparing with the support of GenAI, students achieved statistically significantly better results on both immediate knowledge tests than when preparing in the traditional way. A similar pattern was also observed in the delayed knowledge tests, suggesting that the use of GenAI contributed to greater knowledge retention. Students also reported higher situational interest and greater perceived competence in laboratory work when using GenAI. With regard to attitudes toward GenAI, positive changes were particularly evident in students’ perceptions of the usefulness of GenAI for learning and chemistry instruction. The qualitative findings indicated that students generally perceived GenAI as a fast, accessible, and useful form of learning support. They reported that it helped them understand concepts, check answers, follow step-by-step explanations, and prepare for laboratory work. At the same time, they pointed out several limitations, including the possibility of inaccurate or insufficiently focused responses, the need for appropriate prior knowledge, the difficulty of formulating high-quality prompts, and dilemmas concerning the boundary between acceptable learning support and an inappropriate shortcut. generativna umetna inteligenca kemija laboratorijske vaje dijaki učni dosežki situacijski interes stališča do generativne umetne inteligence generative artificial intelligence chemistry laboratory exercises upper secondary students learning outcomes situational interest attitudes toward generative artificial intelligence. true false false B. Zupančič Slovenski jezik Angleški jezik Magistrsko delo/naloga 2026-07-09 08:31:13 2026-07-09 08:31:22 2026-07-10 11:01:43 0000-00-00 00:00:00 2026 0 Ljubljana 0 Ljubljana VIII, 110, IXVIII str. 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 54:373.3(043.2) 205917 284361987 52203.pdf 52203.pdf 1 3EE5EF552097BA8EC7295376ED82E180 ff8e0c4797438803465bc6aab719133b206e65c7bec97478d3d49952921ddc07 ae767357-7b5f-11f1-9b0d-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=237397 Pedagoška fakulteta 0 0 0