Using Alternative Data and Machine Learning for Predictive Credit Scoring to Promote Financial Inclusion in the U.S.

Abstract

Artificial intelligence is rapidly transforming a wide range of industries in the United States, with the financial sector being one of the most profoundly impacted. One notable advancement is the development of AI driven credit scoring models that use machine learning and large volumes of data to evaluate individual or business credit risk. Unlike traditional credit evaluation methods, which typically rely on financial history, employment records, and credit reports, AI based systems can incorporate alternative data sources such as mobile payment patterns, utility bill records, social media behavior, and even geolocation data. This innovation offers significant potential to expand financial inclusion, especially for underserved communities such as gig workers, rural residents, and individuals with limited credit history. In the United States, where access to conventional financial systems still excludes many due to rigid scoring models, AI offers a more comprehensive view of creditworthiness. However, the growing reliance on algorithmic decision making in finance also raises serious ethical concerns. Biases embedded in historical data or algorithm design may reinforce existing disparities, making it essential to explore the theoretical foundations and real-world implications of AI based credit scoring. This paper examines these emerging opportunities and challenges within the context of the United States financial system.