National Science Foundation AI Institute for Adult Learning and Online Education

• In Event: Cross-Cutting Research Across Three NSF (National Science Foundation )-Funded National AI Institutes on Education

Abstract

Objectives or purposes

The vision of the National AI Research Institute for Adult Learning and Online Education (AI-ALOE) is to use AI to enhance the proficiency of adult learning in online education, making it comparable to that of in-person education. This long-term goal raises significant research issues for both use-inspired and foundational AI.

Perspective(s) or theoretical framework

From the perspective of use-inspired AI, AI-ALOE conducts research on AI cognitive assistants for addressing well-known problems in online education. For example, we are developing intelligent textbooks and smart videos for enhancing cognitive presence, virtual teaching assistants for improving teacher presence, and virtual social assistants for advancing social presence in online education. AI-ALOE’s research on foundational AI focuses on human-AI interaction and personalization of learning. For example, we are developing machine teaching methods for building teachable agents, meta-reasoning techniques for agent self-explanation and information visualization techniques for understanding data. In addition, we are developing techniques for personalization of learning in a variety of learning contexts ranging from skill learning in well-defined problems, to concept learning in ill-defined problems, to model learning in ill-defined and open-ended problems.

Methods, techniques, or modes of inquiry

Personalization of learning in AI assistants can vary from micro-learning where personalization is based on a specific episode of learning using a specific AI tool, to meso-learning where personalization is based on data from multiple data sources and from multiple learning episodes over a class, to macro-learning where personalization in an AI assistant is derived from data over an educational program and from multiple AI tools and data sources. AI-ALOE is developing an architecture for AI-augmented adult learning (A4L) for deploying multiple AI cognitive assistants in online learning contexts; collecting, analyzing, and storing data on learning in standardized formats; and feeding the data to teachers, learners, and the AI assistants. We expect the A4L architecture to support not only micro-learning, but also meso-learning, as well as macro-learning in the long-term.

Data sources, evidence, objects, or materials

We conduct many of our experiments in AI for learning and education at the Technical College System of Georgia which has a very diverse population of adult learners: male and female, majority and minority, urban and rural, civilian and military, and neurotypical and neurodiverse.

Results and/or substantiated conclusions or warrants for arguments/point of view

Early results from our research include personalization of skill learning, personalization of concept learning, machine teaching for constructing teachable agents, meta-reasoning for self-explanation, and integration of cognitive and generative AI for constructing teaching assistants and intelligent textbooks.

Scientific or scholarly significance of the study or work

We expect that research at AI-ALOE will help make adult education simultaneously more available (through online learning and use of online educational materials in blended learning), more affordable (through virtual teaching assistants that offload teachers’ work and amplify their reach), and more achievable (through virtual assistants that support learners cognitively and socially), and thereby, also more equitable.

Authors

• Ashok Goel, Georgia Institute of Technology

• Christopher J. Dede, Harvard University