Learning Paths with Chat

Recent research explores the use of AI-driven adaptive learning platforms to create personalized learning paths for college students. These systems analyze individual student characteristics, pre-existing knowledge, and learning behaviors to customize course content, pacing, and delivery methods (Altaleb et al., 2023; Sharma et al., 2023). Studies have shown that implementing personalized learning paths can significantly improve student performance, engagement, and retention compared to conventional e-learning approaches (Imamah et al., 2024; Jiang et al., 2022). Automated methods, such as ant colony algorithms and cognitive diagnostic assessments, have been developed to generate dynamic learning paths based on student preferences and knowledge point difficulty levels (Imamah et al., 2024; Jiang et al., 2022). These personalized approaches have demonstrated increased course completion rates, learning efficiency, and mastery of knowledge (Jiang et al., 2022). While challenges remain in implementation, the integration of AI-driven personalized learning paths shows promise in addressing diverse learning styles and enhancing educational outcomes (Sharma et al., 2023).

I developed a Business Law course with two learning paths for each of the thirteen substantive course modules. Students were offered a specific learning path based on the student’s score on a pre-test for the specific module. Students that scored under seventy percent on the pre-test were presented one learning path that emphasized lower-level Bloom’s activities (focusing on identifying concepts and defining terms through the use of matching and multiple choice vocabulary activities), where students that scored at least seventy percent were presented an alternate learning path that focused more on the application of concepts (applying concepts from the reading to new scenarios). Both learning paths led to a higher-level analysis or creation activity (such as analyzing a situation in a discussion board activity or synthesizing information gathered to explain how various cases establish a precedent in a particular area of law).

Implementing this approach required extensive development of new activities in the course, primarily for the higher-level Bloom’s learning paths for each module. Chat was extensively used to help develop these new activities, which included the development of LMS-importable items, short writing assignments, contract negotiation scenarios through a Chat-developed web application at https://negotiation.proffaith.com, new Kahoots!, and new discussion activities for the course. In total, I developed approximately thirty new learning activities tied to specific learning objectives throughout the course.

Learning paths were defined using learning management system (LMS) checklists that provided entries with instructions on each activity to be completed by the student in the module. Release conditions were defined on the checklist and individual learning activities based on the pre-test score of the student for that particular module. Generally, a total of five or six learning activities were defined for each learning path, with some activities shared between both paths. Activities within each module were given the same points value, with the lowest scores dropped from each module to address that students on each learning path were completing some different activities within the module. Categories were defined in the LMS grade book utilizing a “weighted” grading system.

At the conclusion of each unit, students were invited to demonstrate their learning through a proctored exam. I then compared the median, standard deviation, and success rates for each of the three exams for control units not exposed to the Learning Paths Model, with treatment units. These results for each exam are summarized below in Table 1.

Table 1 – Comparison of Exam Scores for Learning Paths with Standard Course Design

Notably, the treatment median score on each exam is higher, as is the overall exam pass rate for each of the three exams in the course utilizing the Learning Paths approach. No assessment has been made as to whether the treatment significantly correlates with this outcome on the exams (and there may be other variables that better explain the result), however, these preliminary results are promising and merit further investigation.

Works Cited

Altaleb, Hanan et al. “Enhancing College Education: An AI-driven Adaptive Learning Platform (ALP) for Customized Course Experiences.” 2023 9th International Conference on Optimization and Applications (ICOA) (2023): 1-5. https://doi.org/10.1109/ICOA58279.2023.10308834

Filipe Portela et al. (2024) "Learning Paths: A New Teaching Strategy with Gamification." (2024): 13:1-13:12. https://doi.org/10.4230/OASIcs.ICPEC.2024.13.  

Jiang, B., et al.  (2022) “Data-Driven Personalized Learning Path Planning Based on Cognitive Diagnostic Assessments in MOOCs.” Applied Sciences 12(3982).

Sharma, A.V.N.S., et al. (2024) “Personalized Learning Paths: Adapting Education with AI-Driven Curriculum.” European Economic Letters 14(1). https://doi.org/10.52783/eel.v14i1.993