Promoting AI Trustworthiness through Experiential Learning

Alvis Fong; Steven Carr; Ajay Gupta; Shameek Bhattacharjee

Download Paper | Permalink

Conference: 2022 ASEE Annual Conference & Exposition
Location: Minneapolis, MN
Publication Date: August 23, 2022
Start Date: June 26, 2022
End Date: June 29, 2022
Conference Session: CIT Division Technical Session #10
Page Count: 8
DOI: 10.18260/1-2--40503
Permanent URL: https://peer.asee.org/40503
Download Count: 337

Request a correction

Paper Authors

biography

Alvis Fong Western Michigan University

visit author page

Dr. Fong is with the Department of Computer Science at Western Michigan University. His research interests revolve around data-driven knowledge discovery and aspects of machine intelligence, such as learning for classification and knowledge representation and reasoning. His scientific contributions include two books, fourteen book sections/chapters, two international patents, and 215 papers in reputable journals and conference proceedings. Leading journals that carry his work include IEEE T-KDE, IEEE T-ITBiomed, IEEE T-MM, IEEE T-Evolutionary Computing, IEEE T-Affective Computing, IEEE T-II, and a few other IEEE Transactions titles. He has served on several journal editorial boards and numerous conference committees. Dr. Fong holds four degrees in EE and CS. He is a registered Chartered Engineer and European Engineer.

visit author page

author page

Abstract

Artificial intelligence (AI) systems can sometimes perform undesirably or be manipulated to exhibit biases and abusive behaviors. When AI algorithms are parallelized on high-performance cyberinfrastructure (CI), such misbehaviors and uncertainty can multiply to obscure the root causes. Secure, safe, and reliable (SSR) computing techniques, which are pillars that support AI trustworthiness, can mitigate these problems. This project aims to inform curriculum development by creating and evaluating experiential learning materials to educate STEM students from the outset. Using the educational materials, learners first become aware of the issues and then they develop a range of practical skills toward mitigating them. Intensive, multi-faceted, modular, experiential learning units are designed to rapidly upgrade the skills of current and future CI users, so they become equipped with new skills to apply to their tasks. The loosely coupled modules can be taken as standalone self-directed units to suit CI professionals. The modules can also be integrated into existing classes, starting with CS 1 and CS 2, which are taken by many non-CS STEM students. Three levels of preparedness (foundation, intermediate, and advanced), which roughly correspond to lower undergraduate, upper undergraduate, and graduate, cater to a wide range of learners. In a sandbox environment, learners of each module take measured risks when guided on a challenging yet fun journey of discovery and knowledge acquisition.

Specifically, members of the XXX Research Group at XX University (XXU, anonymized for review), together with public and private partners at various US locations and beyond, aim to address a critical shortage in STEM workforce that understands the anticipated immense technical and societal changes brought about by AI. The need for a strong AI-informed workforce is exemplified by the American AI Initiative. Taking a convergent approach, the project primarily aims to integrate core literacy and advanced skills at the intersection of SSR Computing, High Performance Computing (HPC), and AI into the educational curriculum across multiple STEM disciplines. This will prepare faculty, undergraduate, and graduate students for large-scale data handling and analytics. Our work focuses on institutions that have comparatively lower levels of advanced CI adoption, such as second-tiered institutions (Carnegie Classification R2), historically black colleges and universities (HBCU), and community colleges.

The project’s secondary aim is to lay the groundwork for future broadening adoption of advanced CI training resources that have the potential to influence wide segments of CI communities. This is achieved through identification of best practices derived from the project, modular curricula, and experiential hands-on learning materials. The course is further advanced with carefully designed outreach activities to establish and maintain a pipeline of talents from pre-college to 2-year and 4-year colleges, and graduate programs. Although there is an emphasis on CS curriculum, non-CS STEM students and practitioners who frequently apply AI to their tasks are also intended users of the educational materials.

The purpose of this paper is to share the project team's exciting endeavor broadly. Specific information to be disseminated at the conference include:

1. Design of a series of reproducible, customizable, modular, experiential educational units that can be integrated within existing courses and/or taken as standalone self-directed learning activities. 2. Results (to date) of actual use of the educational units in multiple settings across STEM disciplines (CS, branches of engineering, statistics, business analytics). 3. Proposed outreach activities. 4. Highlights of on-going and future work.

[NSF Division: OAC; Program: CyberTraining]

Citation
Format

Fong, A., & Carr, S., & Gupta, A., & Bhattacharjee, S. (2022, August), Promoting AI Trustworthiness through Experiential Learning Paper presented at 2022 ASEE Annual Conference & Exposition, Minneapolis, MN. 10.18260/1-2--40503

TY - CPAPER
AB - Artificial intelligence (AI) systems can sometimes perform undesirably or be manipulated to exhibit biases and abusive behaviors. When AI algorithms are parallelized on high-performance cyberinfrastructure (CI), such misbehaviors and uncertainty can multiply to obscure the root causes. Secure, safe, and reliable (SSR) computing techniques, which are pillars that support AI trustworthiness, can mitigate these problems. This project aims to inform curriculum development by creating and evaluating experiential learning materials to educate STEM students from the outset. Using the educational materials, learners first become aware of the issues and then they develop a range of practical skills toward mitigating them. Intensive, multi-faceted, modular, experiential learning units are designed to rapidly upgrade the skills of current and future CI users, so they become equipped with new skills to apply to their tasks. The loosely coupled modules can be taken as standalone self-directed units to suit CI professionals. The modules can also be integrated into existing classes, starting with CS 1 and CS 2, which are taken by many non-CS STEM students. Three levels of preparedness (foundation, intermediate, and advanced), which roughly correspond to lower undergraduate, upper undergraduate, and graduate, cater to a wide range of learners. In a sandbox environment, learners of each module take measured risks when guided on a challenging yet fun journey of discovery and knowledge acquisition.

Specifically, members of the XXX Research Group at XX University (XXU, anonymized for review), together with public and private partners at various US locations and beyond, aim to address a critical shortage in STEM workforce that understands the anticipated immense technical and societal changes brought about by AI. The need for a strong AI-informed workforce is exemplified by the American AI Initiative. Taking a convergent approach, the project primarily aims to integrate core literacy and advanced skills at the intersection of SSR Computing, High Performance Computing (HPC), and AI into the educational curriculum across multiple STEM disciplines. This will prepare faculty, undergraduate, and graduate students for large-scale data handling and analytics. Our work focuses on institutions that have comparatively lower levels of advanced CI adoption, such as second-tiered institutions (Carnegie Classification R2), historically black colleges and universities (HBCU), and community colleges.

The project’s secondary aim is to lay the groundwork for future broadening adoption of advanced CI training resources that have the potential to influence wide segments of CI communities. This is achieved through identification of best practices derived from the project, modular curricula, and experiential hands-on learning materials. The course is further advanced with carefully designed outreach activities to establish and maintain a pipeline of talents from pre-college to 2-year and 4-year colleges, and graduate programs. Although there is an emphasis on CS curriculum, non-CS STEM students and practitioners who frequently apply AI to their tasks are also intended users of the educational materials.

The purpose of this paper is to share the project team's exciting endeavor broadly. Specific information to be disseminated at the conference include:

1. Design of a series of reproducible, customizable, modular, experiential educational units that can be integrated within existing courses and/or taken as standalone self-directed learning activities.
2. Results (to date) of actual use of the educational units in multiple settings across STEM disciplines (CS, branches of engineering, statistics, business analytics).
3. Proposed outreach activities.
4. Highlights of on-going and future work.

[NSF Division: OAC; Program: CyberTraining]
AU - Alvis Fong
AU - Steven Carr
AU - Ajay Gupta
AU - Shameek Bhattacharjee
CY - Minneapolis, MN
DA - 2022/08/23
PB - ASEE Conferences
TI - Promoting AI Trustworthiness through Experiential Learning
UR - https://peer.asee.org/40503
DO - 10.18260/1-2--40503
ER -