Search
On-Site Program Calendar
Browse By Day
Browse By Time
Browse By Person
Browse By Room
Browse By Unit
Browse By Session Type
About AERA 2023 Annual Meeting
Program Information
Key Dates / FAQ
Search Tips
Change Preferences / Time Zone
Sign In
Introduction: Cutting-edge educators (Hales et. al., 2019) and researchers (Holstein, McLaren, Aleven, 2019) have certainly explored the use of different consumer-facing AI technologies in their classrooms, but additional work is needed to study student perceptions of these technologies in classrooms (Chang et. al., 2022).
Theoretical Framework: Our commitment to valuing student perceptions is in alignment with a recently published set of recommendations on the use of multimodal learning analytics in ecological learning settings (Author, 2021). Among the authors’ recommendations are guidelines for adopting ethical practices with regard to data collection, data analysis, and data dissemination. Underlying many of their suggestions is a respect for users and their perceptions of how they are being represented through these AI technologies.
Methods and Context: This presentation will look at student perceptions of XXX a collaboration analytics platform that is designed to offer students and instructors data about the nature of their small group collaboration (Author, 2021). XXX provides this data by performing real-time transcription and speech analysis of small group discussions. Aspects of the tool are built on the same technologies found in popular voice assistant devices like Google Home and the Amazon Echo. Different from some of the commercial platforms, however, XXX was designed to include several ethical and privacy-oriented design considerations. For example, the tool, by default, does not store any of the original audio data, nor does it attribute any utterances to a specific student. Instead, the tool provides data at the group level.
Approximately 200 students completed surveys about the XXX platform after having watched a video that described the platform, or after having used the platform for several class sessions. Survey responses were analyzed using a combination of statistical tests and qualitative analysis of free response questions.
Findings: Our analysis of the survey responses suggests that students generally understand the value of the platform, but have differing levels of comfort around its use in the classroom. Students expressed a reluctance to talk during in class discussions when they know that the platform is recording. Other students were content to have their audio data collected, but were reluctant to have video data collected from classroom settings. Still other individuals expressed concern that audio data would not be recorded and stored, because only having access to the automatically generated transcript text would make it hard for students to fix any errors with speech-to-text results.
Discussion: Drawing from these observations and others, the presenters suggest future strategies for thinking about the use of these and similar AI technologies in educational settings. Some of these suggestions are reflected in more recent versions of the platform. For example, the platform is working on including voice distortion capabilities to mask participant identities within audio data. We are also looking at ways to convert video feeds into animations that would preserve certain aspects of people’s identities. The presentation will also include pedagogical considerations around how the technologies are connected to the overall learning experience.