.png){kind=icon}
.png){kind=icon}
Return
Evaluating AI-based visual privacy through Cartoonification: Impact on expressive participation and consent for data retention
.png){kind=icon}
OA
AI
T
K
M
DOI：10.1016/j.caeo.2026.100393.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Learning analytics tools embedded with video devices are increasingly prevalent for capturing student interactions during learning activities. This approach is useful for classroom observation and analyzing students’ emotional expression and perceived confidence during discussions. However, students expressed concern about how their identities could be represented, interpreted, and misused during these analytic and research processes. These privacy concerns are heightened during video-enabled classroom discussions of sensitive topics, as students may feel vulnerable about being recorded for analysis. This study explored the impact of visual privacy through video cartoonification on students’ expressive participation during class discussions on sensitive topics, and their willingness to have their video recorded and used for analysis. Using a multi-method approach, we conducted a study in which students participated in discussions on topics such as police brutality, AI and racial bias, and science-religion tensions, using a platform with an optional cartoonification feature. Data collection included surveys assessing participants' comfort and confidence levels and AI-powered emotional tone analysis of discussion transcripts. The results revealed a strong positive correlation between visual privacy and students' socio-emotional participation levels, reflected in their emotional tone when expressing opinions. Notably, there was a statistically significant difference (p = 0.001) in the expression of concern-emotional tone across all topics when cartoonification was active. Also, this study demonstrates that visual privacy through cartoonification can effectively address students' concerns about visibility in video-enabled learning analytics while maintaining the benefits of capturing student interactions for analysis. The findings provide practical solutions for balancing learning analytics with student privacy concerns.
Keywords:
Visual privacy
Privacy-preserving educational technology
Learning analytics
Emotion detection
Artificial intelligence in education
Sensitive topic discussions
Ethical AI in education
Journal
IF:
5.7
Papers:
373
Citations:
951
Organization
No organization information available
Cited Papers