Preservice Teachers’ Mechanistic Reasoning about Machine Learning and Artificial Intelligence

Amy Voss Farris; Anna Eunji Kim

Download Paper | Permalink

Conference: 2022 Spring ASEE Middle Atlantic Section Conference
Location: Newark, New Jersey
Publication Date: April 22, 2022
Start Date: April 22, 2022
End Date: April 23, 2022
Tagged Topic: Diversity
Page Count: 15
DOI: 10.18260/1-2--40063
Permanent URL: https://peer.asee.org/40063
Download Count: 324

Paper Authors

biography

Amy Voss Farris Pennsylvania State University

visit author page

Amy Voss Farris is currently an Assistant Professor of Science Education at the Pennsylvania State University. She investigates the intersections of scientific modeling and computing in elementary and middle school classrooms and seeks to understand how learners’ and teachers' experiences in scientific computing can support their development of ideas and practices across STEM disciplines. Her teaching encompasses engineering education, preservice teacher preparation, and computational literacies in the Learning Sciences. She is an active member in the International Society of the Learning Sciences (ISLS) and has published numerous journal articles and conference papers on children's scientific and computational modeling in school settings. Her recent work argues that understanding computational thinking requires accounting for the perspectival, material, and embodied experiences in which children’s computing work is grounded.

visit author page

biography

Anna Eunji Kim Pennsylvania State University

visit author page

Anna Eunji Kim is a doctoral student in the Science Education program at Penn State. Her research focuses on the interdisciplinary study of developing the assessment for English language learning students and native English speaking students in Chemistry Education and developing educational curricula within the context of Artificial Intelligence.

visit author page

Download Paper | Permalink

Abstract

Abstract: Modern humans are frequently embedded in contexts in which machines learn from their everyday actions. Examples include encountering predictive text when texting a friend and facial recognition in personal digital photographs. However, explanations that account for the underlying causal mechanisms of machine learning systems require learners to consider parts of the system and the relationships among the parts of that system. While mechanistic reasoning [3] is a foundational body of abilities germane to engineering, children and many adults rarely conceptualize their interactions with machines in ways that are consistent with the complex and dynamic nature of machine learning systems.

We investigated how undergraduate teacher candidates (TCs) explained an example of machine learning, Google Quick, Draw! [13] after playing the game and participating in a series of machine learning investigations. We found that even among an elite group of non-science major undergraduate students, initial explanations of how the computer recognized images rarely focused on how events that lead to a correct guess are linked to one another. In contrast, given opportunities to read and think together about the mechanisms of machine learning, TCs’ descriptions of how Quick, Draw! works became more sophisticated in terms of the sense of mechanism and the chaining [3] of events that lead to a correct guess. For example, in students’ final explanations at the end of our activities, 12 of 22 (54.5%) described the importance of the beginning “stroke” in their doodles or described patterns of key features of doodles as important to how image recognition is accomplished.

Secondly, we explored how building basic understanding of machine learning can support engineering across disciplinary boundaries in K12 contexts. We asked the same preservice teachers to think about how machine learning could be relevant to the content and practices in their area of disciplinary specialization and to create an initial lesson design that could be used with middle school students (U.S. Grades 4 – 8). The participating preservice teachers’ disciplinary specializations were Social Studies (n = 3), English Language Arts (n = 8), and Mathematics (n = 12). We found that all students portrayed that learning goals about artificial intelligence (in general) and machine learning (in particular) were relevant to their focal disciplinary areas and their understanding of literate participation in society. Additionally, some TCs focused on students’ understandings of the social and ethical dimensions of artificial intelligence technologies. This included perceptions of the ethical dimensions of AI and the diverse cultural contexts in which machine learning operates. We report the connections they saw and discuss the relevance of machine learning as an example of reasoning about complex engineered systems for young students and for teachers.

Citation
Format

Farris, A. V., & Kim, A. E. (2022, April), Preservice Teachers’ Mechanistic Reasoning about Machine Learning and Artificial Intelligence Paper presented at 2022 Spring ASEE Middle Atlantic Section Conference, Newark, New Jersey. 10.18260/1-2--40063

TY - CPAPER
AB - Abstract: Modern humans are frequently embedded in contexts in which machines learn from their everyday actions. Examples include encountering predictive text when texting a friend and facial recognition in personal digital photographs. However, explanations that account for the underlying causal mechanisms of machine learning systems require learners to consider parts of the system and the relationships among the parts of that system. While mechanistic reasoning [3] is a foundational body of abilities germane to engineering, children and many adults rarely conceptualize their interactions with machines in ways that are consistent with the complex and dynamic nature of machine learning systems.

We investigated how undergraduate teacher candidates (TCs) explained an example of machine learning, Google Quick, Draw! [13] after playing the game and participating in a series of machine learning investigations. We found that even among an elite group of non-science major undergraduate students, initial explanations of how the computer recognized images rarely focused on how events that lead to a correct guess are linked to one another. In contrast, given opportunities to read and think together about the mechanisms of machine learning, TCs’ descriptions of how Quick, Draw! works became more sophisticated in terms of the sense of mechanism and the chaining [3] of events that lead to a correct guess. For example, in students’ final explanations at the end of our activities, 12 of 22 (54.5%) described the importance of the beginning “stroke” in their doodles or described patterns of key features of doodles as important to how image recognition is accomplished.

Secondly, we explored how building basic understanding of machine learning can support engineering across disciplinary boundaries in K12 contexts. We asked the same preservice teachers to think about how machine learning could be relevant to the content and practices in their area of disciplinary specialization and to create an initial lesson design that could be used with middle school students (U.S. Grades 4 – 8). The participating preservice teachers’ disciplinary specializations were Social Studies (n = 3), English Language Arts (n = 8), and Mathematics (n = 12). We found that all students portrayed that learning goals about artificial intelligence (in general) and machine learning (in particular) were relevant to their focal disciplinary areas and their understanding of literate participation in society. Additionally, some TCs focused on students’ understandings of the social and ethical dimensions of artificial intelligence technologies. This included perceptions of the ethical dimensions of AI and the diverse cultural contexts in which machine learning operates. We report the connections they saw and discuss the relevance of machine learning as an example of reasoning about complex engineered systems for young students and for teachers.

AU - Amy Voss Farris
AU - Anna Eunji Kim
CY - Newark, New Jersey
DA - 2022/04/22
PB - ASEE Conferences
TI - Preservice Teachers’ Mechanistic Reasoning about Machine Learning and Artificial Intelligence
UR - https://peer.asee.org/40063
DO - 10.18260/1-2--40063
ER -