Aesthetic and technical control in AI-generated Animation

Scott Andrew

This project explores the use of AI animation technology in the classroom. During the spring semester, students in an AI generative animation class used tools like Deforum Stable Diffusion and Runway to create both narrative and experimental animations. Their first task was to recreate a past animation made without AI using these new tools. We compared their original and AI-generated animations to assess their technical and aesthetic control. Initially, students had high self-efficacy in non-AI animation but low confidence in AI-based animation. The research indicates that with practice, students' confidence in AI-generated animations matched their non-AI animation confidence. A rubric evaluated their control over aesthetics, technical knowledge, execution, resourcefulness, and project fulfillment. While assessments were somewhat subjective, it was clear that students employed diverse approaches: some created distinct projects conceptually similar to the originals, others enhanced their past work through style transfer, and some aimed for near-identical recreations to test feasibility.

Integrating Ethics through a Transformative Business Education Framework

Rima Bhattacharyay

As seen through the rise of the ESG (environmental social governance) movement and various calls for a more socially conscious capitalism, society and the business community are reckoning with the consequences of marginalizing ethics from the commercial domain. A key actor in responding to the systemic breakdown of business ethics are our universities as they serve as incubators for business professionals. In this capacity, higher level academic institutions are responsible for more than just the transmission of discipline-specific knowledge. They are also responsible for the moral formation of community-minded and socially responsible business leaders. This paper posits a transformative reimagination of how universities and colleges can better serve society and their institutional stakeholders. We explore the unique social opportunity afforded to colleges in re-envisioning the importance of ethics throughout their curricula. Finally, as an example, a case study implemented in the author’s Entrepreneurship business course and the findings from it are explored.

Do students need instruction in using LLMs for writing tasks?

Emily B. DeJeu

Additional author: Walter, Beth, Tepper

Many educators accept that they need to bring LLMs like ChatGPT into the classroom and not ban them. What is less clear is how much instruction, if any, students need in using these tools effectively and ethically. This presentation reports the results of a study that used a pre- and post-course survey as well as a focus group to gather students’ impressions on the extent to which targeted instruction in using LLMs for professional writing proved useful. Results show no meaningful difference in the responses of students who received targeted instruction versus those who did not – students indicated increased familiarity and comfort level with using LLMs for writing tasks regardless of instruction. This pilot study suggests that instructors might be able to integrate LLMs into curricula without giving over too much instructional time to their usage.

Using Generative AI to foster Second Language Learning

Sébastien Dubreil

In an advanced-level French course on oral forms of discourse and storytelling, learners were introduced to generative AI (ChatGPT) as a support for their creativity and writing in a foreign language. Gen AI tools were used in two conditions. In one condition, learners used Gen AI tools to create their initial draft, using AI as a language assistant for support to suggest vocabulary or specific language features (e.g., a rhyme, an alliteration), check the accuracy of sentences, or to edit. In the other condition, learners used Gen AI tools as a creative assistant, adjusting their prompting to create three different "first" drafts that they then refined into a single, final deliverable. Results show evidence that learners in both conditions produced sophisticated texts. In metacognitive reflections they reported both a perceived usefulness to using AI tools and limitations of the tools in expressing their voice. I will also discuss how these results may apply to courses in other fields, especially as regards alignment between student needs and learning objectives.

Using Reflection Assignments to Encourage Engagement with Ethical Theory

Derrick Gray

As part of a Provost’s Inclusive Teaching Fellowship (PITF) project, I worked with Eberly team members to encourage broader and more authentic student engagement with materials and themes covered in Introduction to Ethics, a course in moral philosophy. Here I will discuss the use of reflection exercise assignments, which periodically required students to informally consider the relevance of course topics to their own lives without needing to display mastery of material or related skills. Examples of reflection exercise prompts and anonymized responses will be available, with thoughts on how prompts might be improved to create more authentic engagement. Relatedly, I will also discuss (i) how one might connect the reflection-generated engagement with achievement of other course learning objectives, and (ii) the potential usefulness of reflections in other philosophy classes, including those in which there is less of an obvious connection between topics covered and students’ everyday choices and actions.

Forging Cross-Institutional Research Partnerships: Structure and Agency

Lauren Herckis

Additional authors: Oden Choi, Judeth, Heinz and HCII; Rinsem, Michael, Community College of Allegheny County; Teffera, Leah, HCII; Li, Bruce, Heinz and HCII

This poster identifies barriers to successful cross-sector educational research partnerships and describes effective, evidence-based strategies for overcoming these challenges through a critical case study of one team's efforts. A federally funded project supports collaborative research to develop educational interventions and deploy them at multiple institutions of higher ed. Significant challenges arose in the effort to forge a research partnership between faculty at a research institution and a local community college. Nominally parallel roles in diverse institutions offer widely divergent opportunities, perspectives, engagement with institutional structures, and experiences of academic and personal agency. The research team successfully implemented several strategies to advance research, co-design innovative educational interventions, and implement these in diverse contexts. We share insights gleaned in the course of this collaborative effort which may be useful for others forging cross-institutional research partnerships, particularly between CMU faculty and partners at community colleges or other postsecondary institutions with divergent institutional missions.

AI-mediated Peer Review in the College Writing Classroom

Alan Thomas Kohler

Peer review, the process in which students read each other’s drafts and provide feedback, is seen as foundational to learning and reflection. But, while the benefits of peer review are well-established, students and instructors report that the quality and usefulness of peer feedback–and of the process itself–can often vary wildly. This study investigates how generative AI tools might improve the peer review process in an intermediate undergraduate course where students engaged in both traditional and AI-mediated peer review across five projects. Data were collected through surveys, reflections, and feedback transcripts. Preliminary findings indicate that AI contributed to students' understanding of peer review, but while students found AI-mediated peer review somewhat useful for giving/receiving feedback, they also felt an absence of the human element in the feedback process. This suggests that while generative AI can be a valuable adjunct, it may not substitute for the personal interaction of traditional peer review.

Can you teach creativity? (Yes!)

Amanda Krause

Creativity is a critical skill for engineering students to develop but often ignored in college curricula. This presentation will discuss a research project with the objective to encourage divergent thinking in engineering students. First, in a junior-level class, students were given a creativity lecture that covered the importance of creativity, historical examples, strategies to practice, and the stages of creativity. Then, students were asked to follow the “Creative Stages” to solve engineering problems. The success of this teaching strategy was determined by two assessments: comparisons of pre- and post- surveys regarding self-efficacy and student grades on open-ended problems in class. The surveys revealed that students showed significant growth in their overall self-efficacy for creativity, seeing growth in 13 of the 17 items evaluated. This growth reveals the large opportunity to increase students’ self-efficacy in creativity with small class interventions. Tips for implementation will be discussed.

Facilitating Meaningful and Accessible Robotics Education for Beginners

Alex LaGrassa

Additional authors: Ware, Ranysha, Computer Science Department; Thomas, TJ, Mechanical Engineering

When teaching advanced applications like robot arm programming, the traditional approach starts with a broad set of basic skills before moving to the applications. However, this may not be practical for beginner students who are exploring potential areas of interest. We present a teaching strategy that identifies a meaningful and motivating skill in the application field and then determines the minimal set of skills needed to complete an activity using this skill. This strategy was used in a 2-hour lab-based workshop offered multiple times to students with no programming experience through Black in Robotics. First, we present our learning goals, instructional tools, and examples of meaningful activities that can be performed using five programming concepts. Then, we invite participants to brainstorm how to employ this strategy for a skill they would consider motivating but also too advanced for beginners.

MuFIN: A Framework for Automating Multimodal Feedback Generation using Generative Artificial Intelligence

Jionghao Lin

Additional authors: Chen, Eason, HCII; Gurung, Ashish, HCII; Koedinger, Kenneth R., HCII

Written feedback has long been a cornerstone in educational and professional settings, essential for enhancing learning outcomes. However, multimodal feedback—integrating textual, auditory, and visual cues—promises a more engaging and effective learning experience. By leveraging multiple sensory channels, multimodal feedback better accommodates diverse learning preferences and aids in deeper information retention. Despite its potential, creating multimodal feedback poses challenges, including the need for increased time and resources. Recent advancements in generative artificial intelligence (GenAI) offer solutions to automate the feedback process, predominantly focusing on textual feedback. Yet, the application of GenAI in generating multimodal feedback remains largely unexplored. Our study investigates the use of GenAI techniques to generate multimodal feedback, aiming to provide this feedback for large cohorts of learners, thereby enhancing learning experience and engagement. By exploring the potential of GenAI for this purpose, we propose a framework for automating the generation of multimodal feedback, which we name MuFIN.

The Impact of Generative AI on Student-Generated Microlessons for Learning Sciences

Steven Moore

Additional author: Kwon, Christine

This study examines the impact of generative AI on student-created microlessons in the context of the master’s level course "Tools for Online Learning." We investigated how the integration of ChatGPT in lesson design influences students' lesson quality, understanding of learning science principles, and self-efficacy as educators and AI users. Employing a crossover design, students alternated between using AI support and not across four online modules, each containing two microlesson design tasks. Each task required students to apply core teaching and learning principles, with one microlesson per module designed with AI assistance and one without. By comparing outcomes for each student across different conditions and assignments, the study identifies notable differences in the quality of microlessons, students' learning gains in learning science principles, and changes in their confidence in teaching and using AI.

Optional-Mandatory Attendance

Danny Oppenheimer

Additional authors: Cullen, Simon, Philosophy

Attendance is one of the best predictors of performance in a class, thus to encourage attendance, many teachers make attendance mandatory. While mandatory attendance does increase attendance, it also undermines student motivation and enjoyment of the class. Here we introduce Optional-mandatory attendance - students can choose to opt-in to a mandatory attendance regime (rather than being forced into such a regime by the instructor). We find between 85-100% of students opt in to mandatory attendance, and that this system increases motivation, engagement with course material, and even yields higher attendance than strictly mandatory regimes.

Exploring Student Behavior and Perceived Ability Based on CS Exams

Rhythm Satav

Additional authors: Reese, Abbey, Computer Science; Railing, Brian, Computer Science

This research investigates the effectiveness of randomized computer-based exams used in Carnegie Mellon’s Introduction to Computer Systems (15-213) course. Students receive 7-8 question exams, where questions are chosen randomly among 10-20 others in their distinct categories. The exam system collects data about how each student progresses throughout their exam. This data includes the order in which students view and answer the exam questions, the amount of time they spend on each question viewing, and the score they’ve achieved at any given point during their exam. Our analysis of student scores quantifies and validates the fairness of this exam system. Further analysis of the logs is exploring higher-level student behaviors, such as which question students first solve and their frequency of viewing different questions, and how these behaviors also correlate with ability and score. This research demonstrates the value of administering computer-based exams and informs future exam design.

Enhancing Inclusion in Database Management Education: SQL Teaching for Non-Technical Master’s Programs

Xiaoying Tu

This paper presents a comprehensive revamp of the Introduction to Database Management course at a leading institute for non-technical master's programs to equip them with vital skills owing to the ubiquity of databases across industries. The revamp aims to enhance the relevance of the content by refocusing on SQL, while addressing the goal of inclusive teaching, ensuring students from diverse academic and technical backgrounds can effectively engage with the material. The pedagogical strategies include enhancing in-class instruction techniques, improving transparency in course design, and accommodating diverse backgrounds and learning needs, all aimed at fostering growth mindset and sense of belonging among students. Despite increased content complexity, students achieved high academic standards, increased sense of belonging and maintained robust growth mindset. It contributes to the literature by validating inclusive teaching strategies for non-technical audiences and suggesting their broader applications in university settings to enhance student engagement and success in technical subjects.

PerplexityAI as an Academic Reading Tool for First-Year Students

Liz Walker, Bonnie Youngs

This poster reports on Eberly-sponsored quasi-experimental research in a S24 first-year DC Grand Challenge Seminar. The comparison group was the same course offered in F23. As F23 students experienced difficulties understanding social science and humanities academic articles, our S24 study centered on two questions: (1) how might students design prompts to engage PerplexityAI in support of their reading comprehension?; and (2) how might using PerplexityAI as a reading support tool develop their self-directed learning and critical thinking skills? The research questions focused on and data analysis found: (1) S24 students’ self-efficacy scores on using genAI to read and discuss academic articles improved significantly; and (2) S24 students’ ability to analyze course materials did not exceed that of F23 students. Based on their submitted AI transcripts, we will also discuss what types of questions students asked most often. Our research can apply to courses requiring reading in fields new to students.

Rituals, Reprises, and Chekhov's Emu: On Building a Shared Pedagogical Universe

Gerald J. Wang

In a shared cinematic universe, characters from a common legendarium interact throughout a series of self-contained yet interleaved movies. "Students these days" have grown up in a world where the box office has been consistently dominated by entries from a shared cinematic universe, and so they are particularly attuned to this narrative format. In this roundtable, we discuss strategies for cultivating a shared pedagogical universe, highlighting ideas including the development of shared motifs, traditions, and datasets, both for intra- and inter-semester use. We distill these themes into practices of varying complexity, including several that are particularly easy to implement, and we present examples drawn from a computational science and engineering curriculum. We also provide qualitative and quantitative evidence for the effectiveness of shared pedagogical universes in terms of student learning and engagement.