The project provides an adaptive Intelligent Tutoring System that makes privacy implications graspable by providing students with the option to select their favorite out of 3 versions – differing in the amount of student data collected and, as a result, also the level of personalization. The learning content can be adapted to various ETH courses as a responsible learning tool that teaches students about data literacy and privacy as a secondary learning goal.
The prototype used for the study was based on the tool CTAT – short for Cognitve Tutor Authoring Tools, an ITS developed by Carnegie Mellon University. To account for the varying privacy preferences of students, we created three versions of a prototype with adaptations in data collection and, as a result, the availability of educational features. These three versions were:
Tutor A: This version only records absolutely essential data from students that let the system operate, such as correctness of answers. Also, students could request a hint for a task at hand, but no additional educational features were enabled.
Tutor B: The second version additionally collects and saves data about the development of student’s skills (as defined by the instructor of the course). Also, the hint feature is more customizable in this version, allowing the student to select the type of hint they would like to receive.
Tutor C: Finally, Tutor C additionally contains a tutoring chatbot connected to the task space and therefore has access to the student’s previous interaction with the interface to provide personalized support when asked a question.
