Skip to main content Skip to navigation

"I Would Never Trust Anything Western": Kumu (Educator) Perspectives on Use of LLMs for Culturally Revitalizing CS Education in Hawaiian Schools

Project Overview

The document explores the role of generative AI, specifically large language models (LLMs), in educational contexts, focusing on Indigenous settings like Hawaii's Kaiapuni programs. It reveals that LLMs can significantly enhance efficiency in curriculum development, thereby saving educators time. However, it also identifies critical challenges, including cultural misalignment and reliability issues, underscoring the necessity for AI systems to be tailored to local cultural values and pedagogical practices. The findings advocate for the creation of culturally sensitive AI tools that respect and integrate Indigenous knowledge and educational methodologies, emphasizing that successful implementation of AI in education must prioritize trustworthiness and cultural relevance to foster effective learning environments.

Key Applications

Integration of LLMs in curriculum development for Hawaiian immersion education

Context: Hawaiian public schools and Kaiapuni (immersion language) programs, targeting K-12 educators

Implementation: Surveys and interviews were conducted with kumu (educators) to assess the integration of LLMs in curriculum development.

Outcomes: Time-saving advantages in lesson planning and curriculum design, improved culturally relevant content generation.

Challenges: Cultural misalignment, reliability of generated content, and the need for accurate representation of Hawaiian culture and language.

Implementation Barriers

Cultural Misalignment and Data Reliability

LLMs may produce content that does not accurately reflect or respect Hawaiian cultural values and nuances. There are concerns regarding the accuracy and bias of AI-generated content, particularly for low-resource languages like ‘Ōlelo Hawai‘i.

Proposed Solutions: Develop AI tools with built-in cultural sensitivity features, involve local educators in the design process, and ensure that AI tools provide transparency about their data sources and are trained on culturally appropriate datasets.

Training and Familiarity

Educators lack familiarity and training with AI tools, hindering effective integration into their teaching practices.

Proposed Solutions: Provide comprehensive training resources and user-friendly interfaces for educators.

Project Team

Manas Mhasakar

Researcher

Rachel Baker-Ramos

Researcher

Ben Carter

Researcher

Evyn-Bree Helekahi-Kaiwi

Researcher

Josiah Hester

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Manas Mhasakar, Rachel Baker-Ramos, Ben Carter, Evyn-Bree Helekahi-Kaiwi, Josiah Hester

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies