AI Learning Labs as a Pathway to Productive Use of Energy and Microgrid Feasibility in Off-Grid Communities

Date of Award

5-9-2026

Degree Name

Ph.D. in Mechanical Engineering

Department

Department of Mechanical and Aerospace Engineering

Advisor/Chair

Kevin Hallinan

Abstract

Approximately 775 million people worldwide lack access to any electricity, with sub-Saharan Africa accounting for most of this deficit. Up to 2B people lack consistent access to electricity. While solar microgrids offer a viable electrification pathway, the economic returns necessary to sustain energy investments remain elusive in the absence of a skilled workforce capable of engaging in productive use of energy (PUE) activities. This research identifies education—not energy infrastructure alone—as the critical bottleneck preventing electrified communities from translating energy access into sustainable economic development and proposes artificial intelligence geared toward developing people within a community to use power productively as the mechanism to bridge this gap. This dissertation presents the motivation, design, development, and field validation of an AI-Enabled Adaptive Learning Platform that delivers culturally adapted instruction through conversational AI facilitation. The platform comprises 537 learning modules across seven categories, while also enabling students to create their own learning experiences based upon local interests or challenges, employs a four-level rubric-based AI assessment system, and tracks learner progress through certification pathways aligned with international educational frameworks including UNESCO, ISTE, CSTA, and DigComp 2.2. A companion Community Research Platform automates asset mapping and educational needs assessment through web scraping, AI-facilitated community leader interviews, and Bayesian multi-source data synthesis. The framework was validated through an eight-month (and ongoing) deployment in Oloibiri, Nigeria—a post-oil community where the first computer lab was established in June 2025 with an initial investment of approximately $5,000. Over the deployment period, 46 African-context users generated 1200+ engaged learning sessions and scored assessments, achieving a mean evaluation score of 71.1 with 88.9% of scores at or above the proficiency threshold. Longitudinal chat transcript analysis revealed a 22% increase in student message length, expanded technical vocabulary, and qualitative progression from fragmented definitional queries to structured problem-solving discourse. The deployment also demonstrated a 5.4× engagement disparity between mentor-facilitated and independent users, establishing the human mentor as an indispensable complement to AI-enabled instruction. Community impact included institutional adoption by village leadership, adult spillover usage, peer-to-peer knowledge diffusion, and a waiting list exceeding 1,000 children. These results provide proof-of-concept evidence that AI-enabled conversational education, culturally adapted and mentor-facilitated, can serve as a catalyst for productive use of energy in unpowered communities.

Keywords

Artificial Intelligence, Energy, Engineering, Mechanics

Comments

OCLC No. 1591813509

Rights Statement

Copyright 2026, author.

Link to Full Text

Share

COinS
 
 
 

Links