investigate metaphorical language or uniqueways to describe technical concepts. This can add depth and layers to their poems that they maynot have tried before.This year as a work in progress we decided to try a new path for the students to follow. Theybegan with the instruction to NOT use ChatGPT or any other AI to write their poems. They hadto create what they could and hand it in. The next step was to take that work and put it inChatGPT and create three more versions of their original work. In this paper we explore the useof ChatGPT to not create required work but to show that as a tool ChatGPT opens up the doorsof new forms of creativity, student evaluation of their own work in comparison to the added toolof ChatGPT, and the avenues that a
Paper ID #41739Unfettered ChatGPT Access in First-Year Engineering: Student Usage &PerceptionsDr. Duncan Davis, Northeastern University Duncan Davis is an Associate Teaching Professor in First Year Engineering. His research focuses on using gamification to convey course content in first year classes. He is particularly interested in using the construction of Escape Rooms to teach Engineering Principles.Dr. Nicole Alexandra Batrouny, Northeastern Univeristy Nicole Batrouny is an Assistant Teaching Professor in First Year Engineering at Northeastern University. Her engineering education research interests include the
at the forefront of STEM education experiencing the first tides of thischange. An example of such a trend is the course Design of Machine Elements, a mainstay ofMechanical and Aerospace Engineering (MAE) curricula, which embodies many algorithms thatintegrate a combination of scientific topics and industry protocols. In this work in progress, weassigned a class of 62 MAE machine design students to write computer codes that implementseveral required inputs to generate design parameters for shafts used for specific powertransmission parameters. The students were also asked to explore the applicability of an openartificial intelligence interface, such as ChatGPT, to help develop a multi-step design code. Aftergenerating and verifying the AI
Comparative Analysis of Large Language Models and NLP Algorithms to enhance Student Reflection SummariesAbstractThe advent of state-of-the-art large language models has led to remarkable progress incondensing enormous amounts of information into concise and coherent summaries, benefitingfields like education, health, and public policy, etc. This study contributes to the current effort byinvestigating two NLP approaches’ effectiveness in summarizing students’ reflection text. Thisapproach includes Natural Language Processing (NLP) algorithms customized for summarizingstudents’ reflections and ChatGPT, a state-of-the-art large language model. To conduct the study,we used the CourseMIRROR application to collect students’ reflections from
) have begun to influence software engineeringpractice since the public release of GitHub's Copilot and OpenAI's ChatGPT in 2022. Tools builton LLM technology could revolutionize the way software engineering is practiced, offeringinteractive “assistants” that can answer questions and prototype software. It falls to softwareengineering educators to teach future software engineers how to use such tools well, byincorporating them into their pedagogy.While some institutions have banned ChatGPT, other institutions have opted to issue guidelinesfor its use. Additionally, researchers have proposed strategies to address potential issues in theeducational and professional use of LLMs. As of yet, there have been few studies that report onthe use of LLMs
Boundaries of Engineering Education.AbstractGenerative artificial intelligence (GAI) has long been used across various fields; however, itsusage in engineering education has been limited. Some areas where GAI tools have beenimplemented in education include intelligent tutoring, assessment, predicting, curriculum design,and personalized student learning. The recent proliferation of CHATGPT and other GAI toolspresents limitless possibilities for transforming engineering pedagogy and assessment. At thesame time, there are challenges associated with implementation. Consequently, there is a need toconduct an empirical study to evaluate these tools' strengths, limitations, and challenges tohighlight potential opportunities for their application in
ChatGPT and Google’s Gemini, for the early prediction of studentperformance in STEM education, circumventing the need for extensive data collection orspecialized model training. Utilizing the intrinsic capabilities of these pre-trained LLMs, wedevelop a cost-efficient, training-free strategy for forecasting end-of-semester outcomes based oninitial academic indicators. Our research investigates the efficacy of these LLMs in zero-shotlearning scenarios, focusing on their ability to forecast academic outcomes from minimal input.By incorporating diverse data elements, including students’ background, cognitive, andnon-cognitive factors, we aim to enhance the models’ zero-shot forecasting accuracy. Ourempirical studies on data from first-year college
, 2022, and 2023. A total of 78 students and 3 teachers participated in the program during thistime period.Each team of students submits a project report at the end of the spring semester as part of the programrequirements.3.3 Data Collection Instrument(s)For this study, a total of 10 reports were randomly selected from the participants' submissions. Thesereports were analyzed using Open ChatGPT to explore the students' experiences in the Dual-CreditEngineering program.Open ChatGPT was utilized to conduct a thematic analysis of the reports. Each report was inputted intoOpen ChatGPT, which generated codes based on its content. These codes were then combined to formoverall themes across all 10 reports.The procedure for thematic analysis with Open
objectives on theunderstand level of Bloom’s taxonomy and multiple-choice questions for learning objectives onthe analyze level are shown to moderately achieve this goal. The feedback loop between studentsand instructor was instrumental in determining how to best use class time to support studentlearning. Recommendations for best practices, including how ChatGPT can be leveraged toquickly summarize student responses, based on the instructor’s experience and student feedback,are given.IntroductionStudies have shown that students who read assigned textbook sections before coming to classfind it beneficial for their learning. They have also shown that today’s engineering studentsrarely read the textbook [1]. Just-In-Time-Teaching (JiTT) is a pedagogy
Reshaping Engineering Technology Education: Fostering Critical Thinking through Open-Ended Problems in the Era of Generative AIAbstractAcademic integrity breaches and plagiarism existed long before the rise of Generative Artificialintelligence (G-AI), where students used paid online tutoring platforms like Chegg to obtain helpwith homework assignments, take-home exams, and course projects. Additionally, G-AIplatforms such as ChatGPT provide students with immediate support in understanding conceptsand improving problem-solving abilities. However, it also opens up possibilities for students toimproperly use the technology for homework and exams. This necessitates a revision in howeducators design curricula and
quantitative data.Concurrently, qualitative data was thematically analyzed to gain insights into usage andperceptions surrounding AI.Results: The study revealed a growing trend among project management professionals inleveraging AI tools for a variety of tasks, including project planning, task assignment, tracking,and crafting emails, reports, and presentations. A strong correlation was observed betweenfamiliarity with ChatGPT and its likely usage in project management tasks. While someparticipants found AI tools convenient and efficient, they were frustrated with potentialinaccuracies and the need for specific input prompts. Overall, industry professionalsdemonstrated the usage of AI in project management, with a notable emphasis on taskautomation
technology, but also reported theoutputs generated by the algorithm were not sophisticated enough to be useful for completingcoursework. The question of sophistication is difficult to pin down due to the rapid developmentof the technology, for within the first year of public access, the power of widely availablecommercial platforms like ChatGPT have continued to develop in power and sophistication withthe problems of hallucination and accuracy diminishing as many of the algorithms now haveaccess to the internet, thus further edifying the outputs generated by the AI.Despite these nascent discussions of student impacts, one issue missing from conversationsaround GenAI are the impacts they are likely going to have on how students develop
work.Notably, students who were taught how AI works had significantly different views on AI tools’impact on academic integrity concerns.Computing students’ use of Generative AI is growing, and thoughts on academic integrity are farfrom decided – but there does seem to be an opportunity to teach students the variety of ways itcan be used effectively for programming tasks.IntroductionChatGPT, a Generative AI product developed by OpenAI, was released in November 2022 andalmost immediately, its popularity began to surge worldwide, as illustrated by its steep increaseas a search term on Google. Teachers and administrators took notice – “‘plagiarism’ was rankedin two out of the top five related search queries alongside ‘ChatGPT’” [1]. The popularization
Critical EngagementIn this study, students were invited to participate in a survey to share their experiences using AItools during one semester in four courses. Thirty-five (35) Computer and Electrical Engineering(CEE) students at the University of Wisconsin-Stout responded to the survey describing their useof AI tools such as ChatGPT in their studies. The group included 15 sophomores and 20 seniorsenrolled in 4 different CEE courses titled “CEE-215 Electronics”, “CEE-405 Capstone I:Computer Engineering Design”, “CEE-410 Capstone II: Computer Engineering Design”, and“CEE-355 Applied Electromagnetics”. The survey featured nine questions, seven using a Likertscale to measure students' opinions about AI tools in their education. The Likert scale
, software, andtools can positively impact construction projects by increasing productivity, improving safetyrates, and increasing the success rate of winning construction projects and bids. Interestingly, evenArtificial Intelligence (AI) has made its way into the construction industry, with tools likeChatGPT being utilized to realign project schedules and improve overall project efficiency .Researchers have used ChatGPT to explore integration with digital twins for healthcare, writingmanuscripts, and adapting classroom education to achieve student learning outcomes [18]-[20]. Itis worth noting that tools such as ChatGPT, which have emerged recently as AI-poweredassistants, are still in the process of gathering data to establish their reliability
education (due to the COVID-19 pandemic) as well as a cohort of students whotook all their classes under standard post-pandemic in-person instructional protocols. The secondinterview period also coincided with launch and subsequent public debates around ChatGPT(OpenAI, San Francisco, USA) and other similar generative AI models.All interviews were conducted by the first author virtually using video conferencing. They wereoffered a $50 gift card as a token of gratitude for their time and participation. The interviewsbegan by gathering information about respondents’ educational and employment history andtheir prior training in ethics and public welfare responsibilities. After asking about theirexperiences in their current master’s program, we asked
difficulty • Academic expectations • Learning styles • Assignment deadline • Attendance to class and meeting • Plagiarism, ChatGPT, copying from each other, using a material with a proper or no citation.Some of the challenges faced by students from India are surprising to us because many may notthink that those students may have such challenges in the areas below. • Cultural Adjustment: Many believe India is very close to Western countries because of its unique history over the last 100 years. But we still find that many students adapt to a new culture, lifestyle, and social norms after they arrive in the U.S. Like other international students, they may still experience culture shock, homesickness, and
material. Upon further investigation, it was determined thatwhen asking ChatGPT some specific questions, the responses are also very similar to thatmaterial, suggesting its use of the publisher’s material in its training, and perhaps use of eitherthe materials directly or of ChatGPT by the student development team. In any case, the materialdevelopment phase of the project ended in September, four months before this discovery (inFebruary 2024) so the student development team was unable to support in any corrections thatwere required. The instructor then rewrote and/or restructured many slides prior to use, to ensurethe was no question of copyright infringement.The debugging process proceeded seamlessly, with students in the course finding 29typos
gold standard to evaluateautomated text analytic approaches. Raw text from open-ended questions was converted intonumerical vectors using text vectorization and word embeddings and an unsupervised analysisusing document clustering and topic modeling was performed using LDA and BERT methods. Inaddition to conventional machine learning models, multiple pre-trained open-sourced local LLMswere evaluated (BART and LLaMA) for summarization. The remote online ChatGPTclosed-model services by OpenAI (ChatGPT-3.5 and ChatGPT-4) were excluded due to subjectdata privacy concerns. By comparing the accuracy, recall, and depth of thematic insights derived,we evaluated how effectively the method based on each model categorized and summarizedstudents
were recorded and uploaded on CLAS, they couldsee the difference between their original and improved lessons. It was an empowering learningexperience that gave the preservice teachers the much-needed confidence that they can figurethings out and if a lesson doesn’t go as well as they wanted the first time around, they alwayshave a second chance.Exploration of Novel Pedagogical ApproachesLearning to remove yourself from your own lessons and to reflect on them in order to teachbetter in the future is a core quality of a STEM educator in the 21st century. To be successful inthe era of fast-changing student population, rapidly evolving technologies, that haveunprecedented pedagogical potential, such as ChatGPT [42, 43], continuously
into technical writing instruction.References[1] “Best Practices for Using AI When Writing Scientific Manuscripts: Caution, Care, andConsideration: Creative Science Depends on It” ACS Nano 2023, 17, 5, 4091–4093. 2023.https://doi.org/10.1021/acsnano.3c01544[2] Leung TI, de Azevedo Cardoso T, Mavragani A, Eysenbach G. Best Practices for Using AITools as an Author, Peer Reviewer, or Editor. J Med Internet Res. 2023 Aug 31;25:e51584. doi:10.2196/51584. PMID: 37651164; PMCID: PMC10502596.[3] J. Qadir, "Engineering Education in the Era of ChatGPT: Promise and Pitfalls of GenerativeAI for Education," 2023 IEEE Global Engineering Education Conference (EDUCON), Kuwait,Kuwait, 2023, pp. 1-9, doi: 10.1109/EDUCON54358.2023.10125121.[4] A. Adkins, N. S
eNotebook to include a tutoring AI feature that students could talk to along with their favoritestudy methods. eNotebook provides a general platform for nearly all of today’s study methods andmaterials students use to create and customize for efficient access and assessment. For example,we have implemented a two-way talking conversation feature called Jarvis, which is an audio-to-text / text-to-audio feature with a ChatGPT engine with AI-specific aids to improve the quality ofAI responses. We have embedded weblinks to over 50 of the most popular study apps easilyaccessible through a pull-down menu, where favorites appear at the top of the list. We haveimplemented a feature that converts handwritten notes into typed text. Images, audio, videos
traditionalNLP methods alone [21]. Additionally, as Large Language Models (LLMs) increase and rapidly develop, manyorganizations and researchers compete to create more powerful and advanced GAI models.These new models aim to outperform older versions [22]. GAI models come as applications ortools like ChatGPT, GitHub Copilot, and Bard to name a few. One key example is the GPTmodel, which has gone through versions 3, 3.5, and now 4, each with different capabilities [22].When new GPT versions are released, they often gain new features, capabilities, and parameterscompared to previous versions [22]. Also, OpenAI and other research groups constantly work toimprove LLMs and other AI models. This could impact the accuracy of the information in
can understand what is appropriate for this this task? working well or not in task? the project. Which of the following tools are more appropriate for this problem? Expected Correct Answers Drones Virtual Reality Programming ChatGPT (+1) Point
]. Anotherstudy indicates that ChatGPT-4 outperforms ChatGPT-3.5 and BARD by Google Inc. in several reasoning tasks,particularly in abductive reasoning, mathematical reasoning, and commonsense reasoning [46]. Therefore, in thisstudy, we chose GPT-4 as our preferred LLM model.Educational Implications in Engineering Easy access to psychological monitoring and measurement is imperative in engineering education due to theunique stressors associated with this field. Studies have shown that the engineering culture, often perceived asmasculine, competitive, and exclusionary, can lead to significant stress and mental health challenges for students,particularly for women and students of color [47]. This environment is characterized by a belief in enduring
construction industry, even fewer studies haveexplored the impact of predictive analytics using large language models (LLM), such asChatGPT or BERT. Yet, the little existing research also points to the need for reskilling theexisting workforce [17,18], albeit unclear on the broad implications beyond direct developmentand interaction of LLMs. Short- and long-term implications of the use of ChatGPT in allindustries are still unclear [18]. However, due to its analytical nature, it might displace differentpositions than those affected by the use of robotics.Moreover, it is clear through the increased use of information technologies that roles related tothis discipline will start to be part of the building process. For example, the centrality of
they can right away see being applied through concepts ofsimple Calculus and Python programming.Deep Convolution-based networks with the Triplet loss were quite successful (e.g., FaceNet) inface recognition, resulting in greater than 99% accuracy on benchmarks such as LFW. With therecent success of transformer-based Natural Language Processing architectures (e.g., ChatGPT),transformers have been attempted in Computer Vision applications. They have shown considerablesuccess with better computational efficiency than CNN-based architectures. In this project, wecompared the FaceNet and transformer-based architecture for face recognition. We also providedan insightful understanding of the face recognition process, its limitations, and future
University of Hong Kong, "InnoShow," in Tam Wing Fan Innovation Wing 2023. [Online]. Available: https://innoacademy.engg.hku.hk/innoshow/[24] Innovation Academy, Faculty of Engineering, the University of Hong Kong, "From Ground to Air," in Tam Wing Fan Innovation Wing 2023. [Online]. Available: https://innoacademy.engg.hku.hk/20231106_workshop/[25] Innovation Academy, Faculty of Engineering, the University of Hong Kong, "Build Your IoT Smark Clock," in Tam Wing Fan Innovation Wing 2023. [Online]. Available: https://innoacademy.engg.hku.hk/iotclock/[26] Innovation Academy, Faculty of Engineering, the University of Hong Kong, "Unleash Creativity with Generative AI through Open AI Engine and ChatGPT - Build Your Personalized
the assistance of ChatGPT. We include this information when we share therubric as an incentive for potential collaborators to improve it.) Asset Driven Equitable Partnerships – ADEP in Practice (WIP)References [1] Connor, K. A., & Goodnick, S. M., & Klein, M., & Sullivan, B. J., & Kelly, J. C., & Leigh-Mack, P., & Abraham, S., & Janowiak, J., & Alvarado, S., & Andrei, P., & Scales, W. A., & Wilson, T., & Lagunas, Y. (2023, June), Board 78: ADEP: Asset-Driven Equitable Partnerships (WIP) Paper presented at 2023 ASEE Annual Conference & Exposition, Baltimore , Maryland. 10.18260/1- 2—42939 [2] National Academies of Sciences, Engineering