Conference Session

NSF Grantees Poster Session

Collection

2024 ASEE Annual Conference & Exposition

Authors

Ahatsham Hayat, University of Nebraska, Lincoln; Sharif Wayne Akil, University of Nebraska, Lincoln; Helen Martinez, University of Nebraska, Lincoln; Bilal Khan, Lehigh University; Mohammad Rashedul Hasan, University of Nebraska, Lincoln

Tagged Topics

Diversity, NSF Grantees Poster Session

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

Conference Session

NSF Grantees Poster Session

Collection

2024 ASEE Annual Conference & Exposition

Authors

Yanxia Jia, Arcadia University; Tiantian Wang, The University of Texas at San Antonio; Chaomei Chen, Drexel University; Yu-Fang Jin, The University of Texas at San Antonio

Tagged Topics

NSF Grantees Poster Session

grant data in a CSV format through a converting tool. This feature enables the creationof a network of clusters based on keywords and/or terms (noun phrases) extracted from titles andabstracts of REU awards. Specifically, for each REU award, two different approaches wereadopted to extract terms and keywords. Terms were extracted from the titles and abstracts usingCiteSpace. Technical keyword phrases focusing on research contents of REU awards wereextracted by use of ChatGPT Application Programming Interface (API). Subsequently, anetwork of clusters was created based on the extracted terms and keywords. These clusters revealthe main topics of all REU projects in the dataset.Based on the above-mentioned clusters generated from REU and WoS

Conference Session

NSF Grantees Poster Session

Collection

2024 ASEE Annual Conference & Exposition

Authors

John Clay, University of Texas at Austin; Xingang Li, University of Texas at Austin; Molly H Goldstein, University of Illinois Urbana-Champaign; H. Onan Demirel, Oregon State University; Darya Zabelina; Charles Xie; Zhenghui Sha, University of Texas at Austin

Tagged Topics

NSF Grantees Poster Session

socioeconomic status [16]), whichmay negatively impact design performance. Additionally, the limits of human cognition begin tobe tested as the number and complexity of trade-offs, constraints, and user needs that must beconsidered grows [4], [13]. Finally, traditional/manual design approaches are resource intensivedue to the amount of time required for creating preliminary designs, and for manually correctingpotential errors made by the human designer during these tasks.Figure 1. (a) Genetic algorithms exploring possible solutions for renewable solar-energy systemsin the Aladdin CAD software [8]; (b) Variational autoencoders for structure-aware designgeneration [9]; (c) CAD model generation using large language models, such as ChatGPT [10].Thus

Conference Session

NSF Grantees Poster Session

Collection

2024 ASEE Annual Conference & Exposition

Authors

Ryan Hare, Rowan University

Tagged Topics

NSF Grantees Poster Session

implementation of more automatedsystems in a classroom helps to free up instructor time and resources, and to help raise overallclassroom performance.To achieve an automated educational support system that can stand without instructorintervention, intelligent tutoring systems (ITSs) offer a valuable avenue of research [2]. Thesesystems are well-established in the field, but have seen a surge in development in recent years dueto advancements in large language models like ChatGPT [3], better artificial intelligence methods[4], wider technology adoption, and the recent boom in e-learning [5]. However, a key aspect ofcomputer- or web-based ITSs often remains unaddressed; they are boring.For ITSs to function properly, it is necessary to perform regular

Conference Session

NSF Grantees Poster Session

Collection

2024 ASEE Annual Conference & Exposition

Authors

Grace Lynn Baldwin Kan-uge; Carol S Stwalley P.E., Purdue University, West Lafayette; Robert Merton Stwalley III P.E., Purdue University, West Lafayette

Tagged Topics

NSF Grantees Poster Session

Conference Session

NSF Grantees Poster Session

Collection

2024 ASEE Annual Conference & Exposition

Authors

Corey T Schimpf, University at Buffalo, The State University of New York; Shanna R. Daly, University of Michigan; Leslie Bondaryk, The Concord Consortium; Jutshi Agarwal, University at Buffalo, The State University of New York; Carolyn S Giroux; Stephanie L. Harmon, PIMSER, Eastern Kentucky University; Enqiao (Annie) Fan, University at Buffalo, The State University of New York; Jacqueline Handley, Purdue University, West Lafayette; A Lynn Stephens, The Concord Consortium

Tagged Topics

NSF Grantees Poster Session

of AI techniques and methods toward supporting learning or educational goals.There is a long history of AI being used to support learners from intelligent tutoring systems that trackstudents learning through series of problems and provide custom problem delivery and supports[31], [32],[33] to the more recent use of large-language model, such as ChatGPT, to generate content or support forstudents (e.g., [34]).While AI has been used extensively in some education areas such as math [35], [36], [37] and science[38], [39], it has been used relatively less in design education. Most of the work that does focus on usingAI to support design education tends to examine highly constrained design problems, such as the designof a gear or shaft (e.g., see