From autonomous vehicles to smart personal assistants, artificial intelligence (AI) is increasingly being applied towards solving a wide range of previously intractable problems across multiple disciplines. Taking a convergent approach, the team embarked on a 2.5-year pilot study (Table I). It was an ambitious proof-of-concept investigation aimed at rapidly implementing accessible learning materials to instill the concepts and practice of safe, secure, and reliable (SSR) AI in current and future users of advanced CI. SSR are three pillars that form a sturdy tripod to support trust in AI technologies. Trust in turn is fundamental for enabling effective human-AI collaboration. Harmonious and productive human-AI collaboration has the potential to alter the course of human history, for instance, by enabling scientists and engineers to tackle extremely complex and impactful problems. AI technologies can also profoundly change the ways people live, work, and play. Examples include climate change mitigation, waste reduction and management, delivery of fair and equitable healthcare, transportation, and educational services, clean water and food production, disease prevention, development of smart X (X = cities, factories, infrastructures, ...), robotic pets and companions, detection of harmful content and substances in virtual and physical worlds, etc. 

Following a thorough survey of existing technologies and learning materials, the team supervised several graduate and undergraduate (UG) students and developed 12 initial experiential learning modules (ELMs, summarized in Table II with key characteristics in Table III) in the first project year. The ELMs were designed to complement and leverage existing resources. A panel of academic-industry-government experts provided advice and guidance throughout the design and development process. Some of the experts also subsequently field tested the ELMs in their respective settings during the second launch described below.

In year 2, the team took a 2-step approach for launching the modules. A small-scale launch took place in two CS courses (3000-level UG Big Data and 6000-level grad Information Retrieval) in fall 2021. After minor finetuning, a larger-scale launch occurred across multiple disciplines (e.g., computer science, mechanical engineering, civil engineering, business analytics, and statistics) and institutions in spring 2022. More than 10 faculty at WMU and other institutions took part in the launch, which directly affected more than 300 multidisciplinary UG and grad students. Anonymous data collected in the field test were analyzed by an independent evaluator. The team was engaged in dissemination and outreach activities during the latter half of 2022.

Dissemination and Outreach: Technical (AI biases and mitigation strategies, natural language processing (NLP), commonsense reasoning (CSR), smart health, phishing detection) and learning & teaching papers were published in prominent venues, such as IEEE periodicals and ASEE and ACM SIGITE annual conferences, and SemEval Competitions. Outreach went beyond formal and informal discussions at conferences. For instance, we also worked with area high schools to instill interest in these younger populations. One of the ELMs was customized specifically for outreach activities involving high school students.

Continuity and Forward Momentum: Early in the pilot study, the team uncovered very significant issues surrounding limitations of contemporary statistical machine learning (ML) and AI-induced biases. These limitations and biases hinder effective promotion of AI readiness, and have since become one of the motivating factors that drive further research.

Intellectual Merit: Integrated core and advanced skills at the intersection of SSR computing and AI into educational curriculum to prepare faculty and students at institutions with low rates of advanced cyberinfrastructure (CI) adoption for large-scale data analytics.

Broader Impacts: Several hundred current and future CI users and researchers from diverse backgrounds and locations benefited from the work. Supported 1 graduate and 6 undergraduates (4 minority, 2 female). The second launch directly affected more than 300 graduate and undergraduate students in multiple disciplines. Over ten faculty members from WMU and other institutions were directly involved. In addition to the published papers reported, two presentations were given at prominent teaching & learning conferences. 

Last Modified: 02/04/2023
Modified by: Alvis Fong