2025 ASEE Northeast Section Conference, March 22, 2025, University of Bridgeport, Bridgpeort, CT, USA. Paper-based coding exams: Overcoming coding assessments and grading challenges in the era of AI Large Language Models like ChatGPT Hasan Baig* Phillip Bradford Department of Computer Science Department of Computer Science University of Connecticut University of Connecticut Stamford, CT USA Stamford, CT USA hasan.baig@uconn.edu (*corresponding author
. Misuse swimming pools, retaining walls, domes, dam linings, isof AI can lead to incorrect decision-making. However, instead of challenging. The author was looking for some advanced topicsconsidering AI as proscribed, it can be utilized in research areas to initiate a civil engineering student research project onespecially developing hypothesis and reviewing literature. To application of concrete in complex structures, which was to begenerate a theoretical or experimental research, the researchers started within a week. The author explored ChatGPT [1] to findmust review a series of technical articles. It takes longer if those some suitable topics. At first it was asked that “Can you listarticles are selected
research-based assignments has been exploredless. This study investigates the efficiency and fairness of using AI, specifically ChatGPT, tograde theoretical understanding and research paper assignments in undergraduate and graduatecourses. The research was conducted in two phases. In the first phase, we assessed ChatGPT'sperformance in grading assignments, focusing on time efficiency, consistency, and gradingpatterns. We compared AI-assisted grading with traditional human grading methods in thesecond phase. We then analyzed variations in scores, potential biases, and feedback'sperceived usefulness. We conducted surveys to gather perceptions from both students andeducators regarding AI-based grading.The results indicated that AI-assisted grading
navigate these challenges carefully to educational frameworks represents a significant shift in fully utilize the benefits of AI while preparing students to the way engineering education is approached, with critically engage with the technology (Qadir, 2022). Generative AI tools like ChatGPT leading this The complexity of modern engineering problems transformation. These tools are not only reshaping demands that students not only possess technical curriculum design but are also enhancing the way knowledge, but also advanced problem-solving skills. AI educational content is delivered and interacted with by tools can be instrumental in developing
Analysis (FEA) use AI tools, particularly version of ChatGPT to approximately 500,000 students andChatGPT, to support their learning. A survey examined whether faculty members across 23 campuses [4]. This initiative aimedAI tools help students understand concepts, solve problems to provide personalized tutoring and assist faculty withefficiently, and enhance convenience with 24/7 availability. It also administrative tasks. Similarly, Estonia launched a nationaladdressed their motivational impact, reliability, and concerns initiative in February 2025 to teach AI skills to high schoolabout dependency. The survey statements included questions students, partnering with tech companies like OpenAI
Gen applications in education [1]. In 2022, the public release ofAI (e.g., GPT-4 and DALL-E) have radically expanded AI LLMs like OpenAI’s ChatGPT demonstrated that AI couldcapabilities and access to teaching and learning [2]. Not only engage in sophisticated dialogue, solve complex problems,do these advances enable AI to assess learning data, but they and assist with content generation in real time. Soon after,also generate rich learning content and engage in natural multimodal models emerged that handle not just text but alsolanguage response, blurring the line between human andimagery and other data types, further broadening AI’s capacity enables educational interactions by explaining
itwithin five days, and within two months, it had 100 million active, daily users. Since that time,many generative AI tools have been developed and have been applied to fields ranging from artto medicine, and poetry to finance [1]. Religious groups have even developed AI tools to providechurch-related information and advice, such as Cathy, the Episcopal Church’s “virtual guide.”Cathy (Churchy Answers That Help You) was trained on the Episcopal Church’s website, theBook of Common Prayer and Forward Movement publications, and the ChatGPT knowledgebase. Suggestions for how this bot is to be used include inquiries on official positions of theEpiscopal church, suggestions for liturgy, and general questions about the church and itspractices [4].With
Teaching and Educational Research in EngineeringAbstractThe use of generative Artificial Intelligence (genAI) in teaching and education has receivedattention and rapid growth in university engineering programs since OpenAI released ChatGPT inNovember 2022. In this paper, the authors explore the use of genAI in teaching and educationalresearch in engineering disciplines and examine potential benefits and challenges whiletransitioning to genAI implemented in engineering education. This study A) Analyzes howeducators and learners understand and identify the usage of genAI and ChatGPT in engineeringeducation; B) Explores the potential benefits, challenges, and limitations of using thesetechnologies; and C) Identifies educators' perceptions of using
Engineering Education Randall D. Manteufel Mechanical Aerospace and Industrial Engineering Department University of Texas at San Antonio AbstractSince the introduction of ChatGPT in November 2022, Artificial Intelligence (AI) has been poised tosignificantly impact engineering education by enabling real-time problem-solving assistance,personalized learning experiences, and automated grading systems. The potential uses of AI areextensive, particularly in generating detailed responses to specific queries based on its training data.Ongoing investments and rapid advancements in AI are anticipated to drive breakthroughs
GPTZero and TurnItIn claim to identify whether a student’s writing was One key aspect of this paper is the distinction betweenproprietary and open-source large language models. Proprietary produced by generative AI, but they are highly inaccurate.models, such as OpenAI’s ChatGPT, are often considered less They tend to flag simple or predictable writing as AI-secure and privacy-invasive compared to open-source generated. Studies show that such false positives occur morealternatives like Meta’s Llama. Educating students on the frequently among certain groups, including
and scipy (version 1.7.1) for statistical computations. IEEE Open Journal of the Communications Society, vol. 4, pp. 2952 2971, 2023, doi: 10.1109/OJCOMS.2023.3320646.tools such as Large Language models like ChatGPT are being The sentiment analysis component utilized multiple indicators A moderate positive correlation was observed between self-reported AI [3] Li Y, Wang C, Cao Y, et al. Human pose estimation based in-home lower body rehabilitation system[C]//2020implemented in educational institutions to provide personalized learning
of Blind and Visually Impaired Students and the Impact of Generative AI: A NarrativeAbstractThe advent of Generative AI (GenAI) in our society has taken root so deeply that simple Googlesearches invoke a GenAI response attempting to synthesize a simplified summary for a user.Incidentally, these GenAI systems like ChatGPT from OpenAI, LLaMA from Meta, Geminifrom Google, and Copilot from Microsoft are all largely text-based large language modelsproviding an increased level of access to people who use screen reading technology to interactwith personal computing systems. This study investigates the impact of GenAI on accessibilityfor blind and visually impaired students, focusing on the experiences of two computing
banpersonalized learning and the development of dynamic edu- on ChatGPT by the Italian Data Protection Authority in 2023cational materials. However, the use of GenAI often involvesprocessing sensitive student data, raising concerns about pri- raised concerns about the lack of transparency in AI datavacy and regulatory compliance. This paper examines these collection practices [3]. Institutions must navigate a complexchallenges, highlighting key risks such as data breaches and regulatory landscape, ensuring that GenAI applications alignunauthorized data sharing. A comprehensive solution is proposed with existing legal frameworks such as the GDPR in Europeinvolving privacy-preserving technologies and robust data gov
outputs of bothmodels.For alignment, fuzzy matching techniques were used. These techniques matched sentences be-tween GPT-4o and DeepSeek R1, even when there were minor differences in phrasing. This ap-proach improved the accuracy of mapping and ensured consistency in the processed data. Theresult was a clean and reliable dataset for analysis.5.2. Overall Categorization CoverageWe analyzed the total number of sentences processed and the extent to which GPT-4o and DeepSeekR1 provided category assignments. Table 1 summarizes the categorization coverage across all an-alyzed sentences. ChatGPT DeepSeek Total Sentences 1823 1823
aimed tointegrate artificial intelligence (AI) into the K-12 curriculum by exploring computer vision andAI tools to augment science and technology education. ImageSTEAM specifically introducedvisual media as a critical technology to engage middle school students, particularly in 7th-gradescience, through AI-related topics, digital 3D modeling, and coding.As a result of the workshop, the “Create your 3D Eye” lesson module was developed using AItools such as Pixlr X, TinkerCAD, and ChatGPT prompts. This module helps studentsunderstand the structure and function of the eye and apply their knowledge through interactivedigital tools. The summative assessment for the students is to design and build their 3D model ofan eye from scratch using
opportunities structures in vast datasets as in [1] and continuouslyacross multiple domains, including education, it also raises improving through fine-tuning and prompting.ethical concerns around issues such as copyright,misinformation, and bias. This paper explores the potential of Generative AI operates in three phases:generative AI to revolutionize teaching and learning, with a focuson its impact on student outcomes. Through an examination of • Training, to create a foundation model that can serveselective platforms such as ChatGPT, Character AI, Gemini, and as the basis of multiple gen AI applications.Deep Seek, this paper aims to introduce students to thefundamentals of building AI-powered applications
evolving. In recent years, the development of advancedlarge language models like ChatGPT offer educators powerful tools to enhance teaching practicesand improve classroom experiences. While these tools offer considerable benefits, AI can performtasks such as generating essays, writing code, or solving problems, allowing students to bypassactive learning and rely on the tool to complete their work for them. Previous research hasexplored how instructors and students feel about integrating AI into the classroom as anotherresource. Moving forward, the goal of this study is to build on existing findings and offer newinsights into the perceived benefits and limitations of integrating AI into education by focusing onstudent perceptions of its impact on
artificial intelligence (ai): Understanding the potentialalready known to the respondents, despite the possibility of benefits of chatgpt in promoting teaching and learning,” SSRN, 2023.response (acquiescence) bias. [Online]. Available: https://ssrn.com/abstract=4337484 Finally, the rapid evolution of AI technologies poses chal- [3] P. A. Barrett, A., “Not quite eye to a.i.: student and teacher perspectives on the use of generative artificial intelligence in the writing process,”lenges in capturing a static picture of instructor attitudes. AI Int J Educ Technol High Educ, vol. 20, 2023.models and tools
aid and others seeing it as a risk to independent critical thinking. This study also exploresstudents’ perspectives on integrating AI into future curricula and highlights their suggestions for itsresponsible and effective adoption in engineering education. IntroductionThe rapid advancements in artificial intelligence (AI) are reshaping the education sector. Engineeringeducation has long been at the forefront of adopting technological innovations, reflecting the field'sdynamic and solution-driven nature. AI tools such as ChatGPT, Copilot, Grammarly, Claude,Gemini, Wolfram Alpha are becoming indispensable to enhance learning experiences1,2. Fromautomated routine reminders to facilitating deeper
was assessed through pre/post-Likert-scaleemphasize the importance of strategically designed prompts for surveys (Q2, Q3) and skills assessments.AI tools like ChatGPT, which can foster engagement, critical 2. RQ2 What role does disciplinary context play in shapingthinking, and personalized instruction. The study outlines career relevance and proper perceptions?. This waseffective strategies such as assigning roles to AI, defining clear explored via Likert-scale ratings (Q5) and qualitativeobjectives, and employing iterative dialogue to refine outputs. themes from open-ended responses.Similarly, the authors of [2] explore integrating structuredprompt engineering with generative AI tools. This
the two lists. “Going to ChatGPT helped us create an outline from our B. Positive Aspects of AI use notes to organize the presentation.” Table I presents the profile of responses from Question 1 “Once we searched on kansei engineering + human factors we saw it was a thing and how they combined together.”that explicitly listed perceived advantages of AI. “It was so easy to ask AI to format our references, it saved TABLE I. OVERVIEW OF POSITIVE RESPONES FOR USE OF AI time that was better spent
NTRODUCTION ucators navigating the evolving landscape of AI in education. The integration of artificial intelligence (AI) tools in educa- II. M ETHODOLOGYtion has sparked both enthusiasm and concern among educa-tors and students alike. These tools, ranging from generative A. Course Context and Participants.AI systems like ChatGPT to specialized applications, have the This study was conducted in an undergraduate course titledpotential to reshape how students approach problem-solving, Algorithms and Complexity at the University of Connecticutcollaborate on projects, and prepare for their
learning in higher education.The rapid advancement of these technologies presents both opportunities and challenges foreducators, raising critical questions about the integration of AI into undergraduate classrooms.When systems such as ChatGPT were first introduced, many scholars, such as Noam Chomsky,demonstrated a visceral negative reaction to AI generated text. [1] Generative AI tools were, andlargely still are, seen as a threat to the creative process—ultimately something that academicsshould reject. While these sentiments are perfectly valid, there is a growing body of researchevaluating AI’s benefits. What if there was a way to harness this technology to improve studentengagement and outcomes? Can generative AI personalize learning, automate
our teaching.IntroductionOn Monday, November 4, 2024, during a closed-book test for Statics given on computers, onestudent saw another copy a question, paste it into ChatGPT, and enter the answer just before timeran out on the test.The student cheated, of course, but the argument was made that the test question which askedanything that easily answered by a computer isn’t a good question to ask anymore. For some portionof our teaching careers, it was still good practice to make sure that the students knew the basicsbecause they might not always have a search engine handy. At this time and in the future, it is timeto assume that the students will always be able to look things up easily and quickly.At its heart, education is about preparing our
AbstractThis paper demonstrates the design and implementation of an innovative gamified softwareapplication for learning human-spoken languages. The game serves as an interactive and enjoyablesupplement to aid the learning process of different languages for elementary-aged children. At its core,the application uses a translation Application Programming Interface (API) to process text and outputtranslations in the target language chosen by the learner. Additionally, it is AI-enabled, allowing theutilization of APIs such as OpenAIs’s ChatGPT to enhance the translation capabilities. Provided is abasic proof of concept that was developed as part of the Final Pi Project in the Intermediate ComputerProgramming (COSC 1352) course. The gamified program was
. With the rise of tools like ChatGPT, faculty noticed a perceptible increase inunethical practices resulting in academic dishonesty proliferation throughout the first-yearengineering student population. This necessitated a re-evaluation of assessment methodologies.The first-year engineering cohort of faculty transitioned from the digital to paper format duringSummer 2024 in a small scale, resulting in no academically dishonest behaviors in that smallpopulation. This success positioned the faculty to employ this method of assessment into thestandard procedures of the academic unit.It is important to state the context in which this course of interest is situated. There are a total ofthree courses required by the college of engineering at TAMU
environments. Proceedings of the 2025 ASEE Gulf-Southwest Annual Conference The University of Texas at Arlington, Arlington, TX Copyright © 2025, American Society for Engineering Education 2 IntroductionArtificial Intelligence (AI) has recently influenced the educational landscape. It provides new tools thatenhance teaching efficiency and student engagement. AI-driven applications, such as ChatGPT, areincreasingly used in classrooms to deliver personalized learning experiences, streamline grading, andsupport diverse student needs (Chen et al., 2020). Studies indicate
ChatGPT, Bard, and other generative AI platforms).skills in project and process management. Thoughtfully 1. Transparency - Always disclose when and how you usedesigned AI-integrated assignments not only reinforce course AI tools in your assignments, presentations, or projects.learning objectives but also equip students with the analytical For example, include a brief note in your submissionskills necessary to navigate the future of AI-enhanced stating: "AI tools were used for brainstorming and editingprofessional environments. this document." 2. Responsibility - You are
helpstudents recognize the connections between climate change and engineered systems.Course Format and Generative AI StatementIntroduction to Geotechnical Engineering is a three-credit course which is offered in the fallsemester that meets three times a week (50 minutes each). The course focuses on engineering useof soils; lab and field determination of soil properties; determination of phase relationships; soilclassification; soil-water interaction; stress effects of loading on soils at depth; consolidation,compaction, shear strength, bearing capacity theory, and several special geotechnical topics.Certain assignments in this course may allow the use of generative artificial intelligence (AI)tools such as ChatGPT. The default is that such use is
intersecting factors on theaccessibility of educational resources, opportunities, accommodations, and support systems.In recent years, the pursuit of educational equity has increasingly intersected with advancementsin technology, particularly artificial intelligence (AI). Just as earlier legal and policy reformssought to address the systemic barriers faced by marginalized groups, technological innovationsare opening new pathways to equitable education. A pivotal moment in AI research occurred inMarch 2016, when AlphaGo defeated the world chess champion, capturing global attention andsparking global interest across numerous fields. In education, AI-driven tools have similarlyushered in a new era, with tools like ChatGPT. Introduced in November 2022