JPMorgan Chase developing system to create fair synthetic data for AI training

From Yahoo Finance: 2025-06-09 11:15:00

JPMorgan Chase is seeking to patent a system for generating fair synthetic representative data to train machine learning models. The system aims to create synthetic data that mimics real data in statistical accuracy, demographic information, and demographic parity to prevent unfair bias. This move highlights the importance of mitigating biases in AI models to ensure equity.

Chase’s system collects demographic information, decision-making features, and outcomes to create a synthetic dataset that closely resembles the original data. It then adjusts the weights of the synthetic data to ensure demographic parity. The company’s focus on AI reflects its commitment to integrating technology to improve efficiency and reduce headcount in operations and account services.

The development of a system to generate fair synthetic data is crucial for financial institutions like Chase to prevent bias in AI models, particularly in tasks like loan approvals where historical bias has disadvantaged certain groups. By tackling bias in data, Chase can advance its AI goals while minimizing risks associated with biased decision-making.



Read more at Yahoo Finance: JPMorgan Chase Patent Tackles Synthetic Data Bias