Learning to Adopt Generative AI
Project Overview
The document examines the influence of generative AI, particularly ChatGPT, in education, emphasizing the disparities in its adoption and effectiveness among various demographic groups. It introduces the concepts of the 'learning divide,' which describes the differing capabilities of individuals to adapt their understanding of AI's usefulness, and the 'utility divide,' highlighting the variations in the tangible benefits received from AI tools. The findings indicate that while certain demographic groups may experience greater advantages from generative AI, their slower rates of learning impede effective utilization. To address these disparities, the study advocates for the implementation of targeted training programs aimed at promoting equitable access to AI resources and improving overall educational outcomes.
Key Applications
ChatGPT
Context: Individuals from various demographic backgrounds using generative AI for educational purposes
Implementation: Utilization of a Bayesian learning model to analyze user interactions with ChatGPT
Outcomes: Identification of learning and utility divides based on demographic attributes; users with lower education levels derive higher utility but learn more slowly
Challenges: Slower learning rates among lower-educated and non-white users can lead to underutilization and belief traps
Implementation Barriers
Disparities in Generative AI Utilization
Includes the Learning Divide, which refers to disparities in individuals' abilities to effectively update their perceived utility of generative AI through repeated interactions, and the Utility Divide, which highlights differences in the actual utility derived from generative AI among individuals from different demographic backgrounds.
Proposed Solutions: Implementing training programs aimed at enhancing exposure and belief updating processes, alongside inclusive design practices and targeted user training to ensure equitable access to AI benefits.
Project Team
Lijia Ma
Researcher
Xingchen Xu
Researcher
Yumei He
Researcher
Yong Tan
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Lijia Ma, Xingchen Xu, Yumei He, Yong Tan
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