ChatGPT, Copilot, Gemini, SciSpace and Wolfram versus higher education assessments: an updated multi-institutional study of the academic integrity impacts of Generative Artificial Intelligence (GenAI) on assessment, teaching and learning in engineering

Australasian journal of engineering education, Год журнала: 2024, Номер unknown, С. 1 - 28

Опубликована: Июль 11, 2024

More than a year has passed since reports of ChatGPT-3.5's capability to pass exams sent shockwaves through education circles. These initial concerns led multi-institutional and multi-disciplinary study assess the performance Generative Artificial Intelligence (GenAI) against assessment tasks used across 10 engineering subjects, showcasing GenAI. Assessment types included online quiz, numerical, oral, visual, programming writing (experimentation, project, reflection critical thinking, research). Twelve months later, was repeated using new updated tools ChatGPT-4, Copilot, Gemini, SciSpace Wolfram. The investigated differences, identifying best tool for each type. findings show that increased features can only heighten academic integrity concerns. While cheating are central, opportunities integrate GenAI enhance teaching learning possible. had specific strengths weaknesses, ChatGPT-4 well-rounded. A Security Opportunity Matrix is presented provide community practical guidance on managing risks integration learning.

Язык: Английский

---

Need a Guide by My Side Not a Sage on Stage: ChatGPT in Learning

Lecture notes in networks and systems, Год журнала: 2025, Номер unknown, С. 135 - 145

Опубликована: Янв. 1, 2025

Язык: Английский

---

The role of generative AI tools in shaping mechanical engineering education from an undergraduate perspective

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 17, 2025

Abstract This study evaluates the effectiveness of three leading generative AI tools-ChatGPT, Gemini, and Copilot-in undergraduate mechanical engineering education using a mixed-methods approach. The performance these tools was assessed on 800 questions spanning seven core subjects, covering multiple-choice, numerical, theory-based formats. While all demonstrated strong in questions, they struggled with numerical problem-solving, particularly areas requiring deep conceptual understanding complex calculations. Among them, Copilot achieved highest accuracy (60.38%), followed by Gemini (57.13%) ChatGPT (46.63%). To complement findings, survey 172 students interviews 20 participants provided insights into user experiences, challenges, perceptions academic settings. Thematic analysis revealed concerns regarding AI’s reliability tasks its potential impact students’ problem-solving abilities. Based results, this offers strategic recommendations for integrating curricula, ensuring responsible use to enhance learning without fostering dependency. Additionally, we propose instructional strategies help educators adapt assessment methods era AI-assisted learning. These findings contribute broader discussion role implications future methodologies.

Язык: Английский

---

ChatGPT‐3.5 and ‐4.0 and mechanical engineering: Examining performance on the FE mechanical engineering and undergraduate exams

Computer Applications in Engineering Education, Год журнала: 2024, Номер 32(6)

Опубликована: Июль 14, 2024

Abstract The launch of Generative Pretrained Transformer (ChatGPT) at the end 2022 generated large interest in possible applications artificial intelligence (AI) science, technology, engineering, and mathematics (STEM) education among STEM professions. As a result many questions surrounding capabilities generative AI tools inside outside classroom have been raised are starting to be explored. This study examines ChatGPT within discipline mechanical engineering. It aims examine use cases pitfalls such technology professional settings. was presented with set from junior‐ senior‐level engineering exams provided private university, as well practice for Fundamentals Engineering (FE) exam responses two models, one free paid subscription, were analyzed. paper found that subscription model (GPT‐4, May 12, 2023) greatly outperformed version (GPT‐3.5, 2023), achieving 76% correct versus 51% correct, but limitation text only input on both models makes neither likely pass FE exam. results confirm findings literature regard types errors made by ChatGPT. due its inconsistency tendency confidently produce incorrect answers, tool is best suited users expert knowledge.

Язык: Английский

---

Teachers' perceptions on the introduction of Generative AI in schools: A mixed-method study on the opinions of 1,223 teachers in the Veneto Region, Italy

EDUCATION SCIENCES AND SOCIETY, Год журнала: 2025, Номер 2, С. 17 - 37

Опубликована: Янв. 1, 2025

Generative Artificial Intelligence (GenAI) is gaining momentum in schools as a means of support to the teaching and learning process. However, its use poses several controversial questions, especially lower school grades, teachers might often face ethical or intellectual obstacles preventing them from using AI their classes. This study explores perceptions sample 1,223 across subjects instruction 572 regional context (nursery, primary, upper secondary), mixed-method approach. Results suggest that there widespread confusion on possible applications GenAI education, possibly leading reduced teachers' intention integrate these tools practices. also point towards general need for more CPD topic. Age, level subject were found moderate effect perceived readiness GenAI. Regarding negative implementations GenAI, showed have mixed opinions, open contrast unreserved enthusiasm. Limitations future research lines are addressed.

Язык: Английский

---

Transforming Education

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 89 - 102

Опубликована: Март 13, 2025

This chapter delves into the transformative potential of prompt engineering for enhancing interactive learning by integrating advanced GenAI technologies. The strategic employment offers a unique opportunity to customize and enrich environments. Educators can design engaging, adaptive, personalized educational experiences harnessing different AI tools. It examines how specific instructions iterative processes involved in optimize AI-generated outputs improve quality content. Through various case studies, from elementary professional training, effectiveness tailored prompts fostering deeper understanding, critical thinking, creative problem-solving is highlighted. challenges ethical considerations are explored ensure balanced perspective on risks rewards. guides stakeholders utilizing foster more interactive, responsive, inclusive landscape.

Язык: Английский

---

Educational innovation: Exploring the Potential of Generative Artificial Intelligence in cognitive schema building

Edutec Revista Electrónica de Tecnología Educativa, Год журнала: 2024, Номер 89, С. 44 - 63

Опубликована: Сен. 30, 2024

This study explores the use of generative artificial intelligence to enhance teaching and learning experience, focusing on strengthening consolidating cognitive schemas. Research reveals that schemas can profoundly influence improvement experience promote assimilation new types information retention in students' memory. To improve advantages, obstacles, potential future trajectories utilizing these technologies were examined by conducting a thorough literature review analyzing relevant studies. Findings indicate has personalize learning, diversify educational content, efficiency scalability. However, it also poses challenges related content quality, data privacy, equity access personalized learning. Future research should focus effectiveness tools based AI inclusion, ethical approaches, interdisciplinary collaboration. Overall, this provides solid foundation for understanding harnessing enhancing schemas, thereby promoting more effective, inclusive, education.

Язык: Английский

---

Why students use or not use generative AI: Student conceptions, concerns, and implications for engineering education

Digital engineering., Год журнала: 2024, Номер unknown, С. 100019 - 100019

Опубликована: Ноя. 1, 2024

Язык: Английский

---

Bard, ChatGPT and 3DGPT: a scientometric analysis of generative AI tools and assessment of implications for mechanical engineering education

Interactive Technology and Smart Education, Год журнала: 2024, Номер 21(4), С. 588 - 624

Опубликована: Фев. 14, 2024

Purpose Following the recent rise in generative artificial intelligence (GenAI) tools, fundamental questions about their wider impacts have started to reverberate around various disciplines. This study aims track unfolding landscape of general issues surrounding GenAI tools and elucidate specific opportunities limitations these as part technology-assisted enhancement mechanical engineering education professional practices. Design/methodology/approach As investigation, authors conduct present a brief scientometric analysis recently published studies unravel emerging trend on subject matter. Furthermore, experimentation was done with selected (Bard, ChatGPT, DALL.E 3DGPT) for engineering-related tasks. Findings The identified several pedagogical guidelines deploying engineering. Besides, highlights some pitfalls analytical reasoning tasks (e.g., subtle errors computation involving unit conversions) sketching/image generation poor demonstration symmetry). Originality/value To best authors’ knowledge, this presents first thorough assessment potential from lens field. Combining analysis, insights, provides unique focus implications material selection/discovery product design, manufacturing troubleshooting, technical documentation positioning, among others.

Язык: Английский

---

A Data-Driven Approach for the Identification of Features for Automated Feedback on Academic Essays

IEEE Transactions on Learning Technologies, Год журнала: 2023, Номер 16(6), С. 914 - 925

Опубликована: Сен. 29, 2023

For predicting and improving the quality of essays, text analytic metrics (surface, syntactic, morphological, semantic features) can be used to provide formative feedback students in higher education. In this study, goal was identify a sufficient number features that exhibit fair proxy scores given by human raters via data-driven approach. Using an existing corpus analysis tool for Dutch language, large were extracted. Artificial neural networks, Levenberg–Marquardt algorithm, backward elimination reduce automatically. Irrelevant eliminated based on inter-rater agreement between predicted calculated using Cohen's kappa ( $\kappa$ ). The study reduced from 457 28 grouped into different categories. results reported article are improvement over similar previous study. First, reliability increased tweaking overfitting average scores. resulting maximum value showed substantial compared moderate prior Second, instead dedicated training test set, testing phases new experiments performed notation="LaTeX">$k$ -fold cross validation texts. approach presented research is first step toward our ultimate providing meaningful enhancing their writing skills capabilities.

Язык: Английский

---