Опубликована: Сен. 24, 2024
Язык: Английский
Frontiers in artificial intelligence and applications, Год журнала: 2024, Номер unknown
Опубликована: Июнь 5, 2024
The general availability of large language models and thus unrestricted usage in sensitive areas everyday life, such as education, remains a major debate. We argue that employing generative artificial intelligence (AI) tools warrants informed examined their impact on problem solving strategies higher education. In study, students with background physics were assigned to solve exercises, one group having access an internet search engine (N=12) the other being allowed use ChatGPT (N=27). evaluated performance, strategies, interaction provided tools. Our results showed nearly half solutions support mistakenly assumed be correct by students, indicating they overly trusted even field expertise. Likewise, 42% cases, used copy & paste query — approach only 4% queries highlighting stark differences behavior between groups limited task reflection when using ChatGPT. our work, we demonstrated need (1) guide how interact LLMs (2) create awareness potential shortcomings for users.
Язык: Английский
Процитировано
17
European Journal of Physics, Год журнала: 2023, Номер 45(2), С. 025701 - 025701
Опубликована: Дек. 11, 2023
Abstract The paper aims to fulfil three main functions: (1) serve as an introduction for the physics education community functioning of large language models (LLMs), (2) present a series illustrative examples demonstrating how prompt-engineering techniques can impact LLMs performance on conceptual tasks and (3) discuss potential implications understanding prompt engineering teaching learning. We first summarise existing research popular LLM-based chatbot (ChatGPT) tasks. then give basic account work, illustrate essential features their functioning, strengths limitations. Equipped with this knowledge, we some challenges generating useful output ChatGPT-4 in context introductory physics, paying special attention questions problems. provide condensed overview relevant literature demonstrate through selected be employed improve ’s Qualitatively studying these provides additional insights into ChatGPT’s its utility problem-solving. Finally, consider from inform use learning physics.
Язык: Английский
Процитировано
37
Trends in Higher Education, Год журнала: 2025, Номер 4(1), С. 2 - 2
Опубликована: Янв. 8, 2025
This collective systematic literature review is part of an Erasmus+ project, “TaLAI: Teaching and Learning with AI in Higher Education”. The investigates the current state Generative Artificial Intelligence (GenAI) higher education, aiming to inform curriculum design further developments within digital education. Employing a descriptive, textual narrative synthesis approach, study analysed across four thematic areas: learning objectives, teaching activities, development, institutional support for ethical responsible GenAI use. 93 peer-reviewed articles from eight databases using keyword-based search strategy, collaborative coding process involving multiple researchers, vivo transparent documentation. findings provide overview recommendations integrating into learning, contributing development effective AI-enhanced environments reveals consensus on importance incorporating Common themes like mentorship, personalised creativity, emotional intelligence, higher-order thinking highlight persistent need align human-centred educational practices capabilities technologies.
Язык: Английский
Процитировано
1
International Journal of Educational Technology in Higher Education, Год журнала: 2025, Номер 22(1)
Опубликована: Фев. 9, 2025
Abstract This paper presents a systematic review of the role prompt engineering during interactions with Generative Artificial Intelligence (GenAI) in Higher Education (HE) to discover potential methods improving educational outcomes. Drawing on comprehensive search academic databases and relevant literature, key trends, including multiple framework designs, are presented explored role, relevance, applicability purposefully improve GenAI-generated responses higher education contexts. Multiple experiments using variety frameworks compared, contrasted discussed. Analysis reveals that well-designed prompts have transform GenAI teaching learning. Further findings show it is important develop teach pragmatic skills AI interaction, meaningful engineering, which best managed through for creating evaluating applications aligned pre-determined contextual goals. The outlines some concepts educators should be aware when incorporating into their practices, students necessary successful interaction.
Язык: Английский
Процитировано
1
Journal of Mechanical Design, Год журнала: 2024, Номер 146(11)
Опубликована: Май 8, 2024
Abstract Conceptual design is a pivotal phase of product and development, encompassing user requirement exploration informed solution generation. Recent generative models with their powerful content generation capabilities have been applied to conceptual support designers’ ideation. However, the lack transparency in process shallow nature generated solutions constrain performance complex tasks. In this study, we first introduce approach that combines classic theory. This decomposes task based on attributes, uses who, what, where, when, why, how (5W1H) method, function-behavior-structure model, Kansei Engineering guide generate through multi-step reasoning. Then present an interactive system using mind-map layout visualize reasoning, called DesignFusion. empowers designers track control inputs/outputs at each reasoning step. Two studies show our significantly enhances quality enriches designer experience human–artificial intelligence co-creation.
Язык: Английский
Процитировано
5
Physical Review Physics Education Research, Год журнала: 2024, Номер 20(1)
Опубликована: Май 23, 2024
As a result of the pandemic, many physics courses moved online. Alongside, popularity Internet-based problem-solving sites and forums rose. With emergence large language models, another shift occurred. One year into public availability these how has online help-seeking behavior among introductory students changed, what is effect different patterns resource usage? In mixed-method approach, we investigate student choices their impact on assessment components an course for scientists engineers. We find that still mostly rely traditional Internet resources usage strongly influences outcome low-stake unsupervised quizzes. empirically found distinct clusters resource-usage students; students’ cluster membership supervised course, however, nonsignificant. Published by American Physical Society 2024
Язык: Английский
Процитировано
5
Journal of Baltic Science Education, Год журнала: 2025, Номер 24(1), С. 187 - 207
Опубликована: Фев. 25, 2025
As the development and application of large language models (LLMs) in physics education progress, well-known AI-based chatbot ChatGPT4 has presented numerous opportunities for educational assessment. Investigating potential AI tools practical assessment carries profound significance. This study explored comparative performance human graders scoring upper-secondary essay questions. Eighty students’ responses to two questions were evaluated by 30 pre-service teachers ChatGPT4. The analysis highlighted their consistency accuracy, including intra-human comparisons, GPT grading at different times, human-GPT variations across cognitive categories. intraclass correlation coefficient (ICC) was used assess consistency, while accuracy illustrated through Pearson with expert scores. findings reveal that demonstrated higher scoring, scorers showed superior most instances. These results underscore strengths limitations using LLMs assessments. high can be valuable standardizing assessments diverse contexts, nuanced understanding flexibility are irreplaceable handling complex subjective evaluations. Keywords: Physics question assessment, grader, Human graders.
Язык: Английский
Процитировано
0
Osmaniye Korkut Ata Üniversitesi Fen Bilimleri Enstitüsü Dergisi, Год журнала: 2025, Номер 8(2), С. 598 - 609
Опубликована: Март 12, 2025
This study examines the impact of one AI conversational tools, ChatGPT-4, on learning outcomes in construction management courses. Focusing Site Management course, same exam questions were administered to both students and ChatGPT-4. evaluation was conducted two phases—midterm final exams—where performances ChatGPT-4 compared analyzed. The research reveals that while outperformed midterm exam, it showed a decline performance during indicating limitations adapting different environments. Neither nor met course passing criteria. emphasizes need for further development more effective integration into education, with focus considering ethical dimensions this process.
Язык: Английский
Процитировано
0
Lecture notes in networks and systems, Год журнала: 2025, Номер unknown, С. 220 - 227
Опубликована: Янв. 1, 2025
Язык: Английский
Процитировано
0
Lecture notes in networks and systems, Год журнала: 2025, Номер unknown, С. 141 - 148
Опубликована: Янв. 1, 2025
Язык: Английский
Процитировано
0