MIT researchers develop AI education program for U.S. Air Force and Department of Defense
The work was published in a peer-reviewed study that analyzed a pilot educational program for learners with diverse backgrounds
A team of researchers at MIT Open Learning and the MIT Media Lab developed an academic program to teach U.S. Air Force (USAF) personnel to understand and utilize artificial intelligence technologies.
The work was recently published in an IEEE paper (part of the IEEE Frontiers in Education Conference) titled “Designing and implementing an artificial intelligence (AI) education program for learners with diverse background at scale.” According to MIT News, the program was found to be “effective and well-received by employees with diverse backgrounds and professional roles.”
The paper was co-authored by Andres F. Salazar-Gomez, Aikaterini Bagiati, Nicholas Minicucci, Kathleen D. Kennedy, Xiaoxue Du, and Cynthia Breazeal PhD ’00. Du is a researcher at the Media Lab and Salazar-Gomez, Bagiati, Minicucci, and Kennedy are researchers at Open Learning. Brezeal, who serves as MIT’s Dean for Digital Learning, director of MIT RAISE (Responsible AI for Social Empowerment and Education), and head of the Media Lab’s Personal Robots research group, was the principal investigator for the study.
MIT entered into an agreement with the USAF and the Department of Defense (DoD) to develop the program in January of 2021, with the goal of eventually creating a program that could be used for people of backgrounds ranging from the general public to DoD personnel. The paper focuses on the results of the first pilot; these results will inform additional iterations.
According to the paper, the study analyzed the effectiveness of various aspects of the educational experience — including AI content and curriculum, pedagogical approaches, learning modalities, and technological innovations.
The study attempted to create AI curricula to address six different “learner profiles” and their desired AI skills and competencies; these profiles were Lead AI, learners who would want to eventually make policy decisions about AI tools; Drive AI, learners who would need to ensure that appropriate AI tools and capabilities are developed; Create AI, learners who want to develop AI tools; Embed AI, learners who would establish AI systems and tactics; Facilitate AI, learners who delivered to particular use cases; and Employ AI, learners who would use the AI tools. These profiles were combined into cohorts named Lead-Drive, Create-Embed, and Facilitate-Employ.
The curricula for the three cohorts were developed in the summer of 2020; along with the curricula, a combination of different learning modalities was selected for each cohort, employing self-paced and instructor-led asynchronous and synchronous courses, webinars, and experiential workshops. The curricula consisted of foundational concepts, applications, data visualization, responsible AI, coding, statistics, and AI delivery and enablement, with each cohort’s program differing in the depth to which each of these areas was studied.
The data came from several surveys filled out by the learners – USAF and DoD personnel invited to participate in the study. The initial data analysis focused on survey participation, content difficulty, program completion time, pedagogy comfort level, technology usability, interest in the content, and relevance with current USAF/DoD work. The surveys also solicited recommendations for improvement.
The researchers noted that learner participation in the study decreased with time; 78%, 76%, and 63% of participants completed the program from the Lead-Drive, Created-Embed, and Facilitate-Employ cohorts respectively.
The Lead-Drive cohort responded positively to the experiential, hands-on learning, while members of the Facilitate-Employ cohort responded negatively to the lack of live communication. Members of all cohorts requested more case studies and examples of applications of AI with relevance to the USAF and DoD.
Brezeal said to MIT News that “We are digging deeper into expanding what we think the opportunities for learning are, that are driven by our research questions but also from understanding the science of learning about this kind of scale and complexity of a project.”
“We are also trying to deliver some real translational value to the Air Force and the Department of Defense. This work is leading to a real-world impact for them, and that is really exciting,” Brezeal concluded.