Generative AI for adaptive tutoring and college student success
Description: In this talk, I'll describe results from a series of empirical studies evaluating the ability of current LLMs to generate questions with similar psychometric properties to textbook questions, generate hints with similar learning gains to human-authored hints, and conduct curricular...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Language: | English |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/181113 https://www.ntu.edu.sg/mae/ai-education-singapore-2024/activities/keynote-invited-talk#Content_C021_Col00 |
| Summary: | Description: In this talk, I'll describe results from a series of empirical studies evaluating the ability of current LLMs to generate questions with similar psychometric properties to textbook questions, generate hints with similar learning gains to human-authored hints, and conduct curricular alignment of GenAI educational resources to existing taxonomies and syllabi. These publications, out of the Computational Approaches to Human Learning research lab at the UC Berkeley School of Education move the field closer to automatically generated, mastery-based, Intelligent Tutoring Systems and build upon an existing open source and creative commons project, called Open Adaptive Tutor (OATutor). I will also discuss how the same LLM technology is finding equivalencies in college curricula, allowing for new frontiers in credit mobility to be paved across large public systems of higher education. |
|---|