PeopleFund Develops Its Own 'Explainable AI Model'

Multidimensional Explanation of the Basis and Reasons for AI Credit Evaluation Model Outputs

PeopleFund (CEO Kim Dae-yoon) announced on the 29th that it has independently developed and commercialized four types of "Explainable Artificial Intelligence (AI) models (XAI models)" to be used in the fields of credit screening and evaluation. PeopleFund's AI credit evaluation system is said to have established an objective and transparent AI explanation framework.

XAI is a methodology that enables users to understand and trust the operating principles of AI and the basis of its decision-making. PeopleFund's XAI models explain the rationale behind borrower evaluations in the AI credit evaluation system and include functions that assess and explain whether the credit evaluation models in use are adapting well to changes in market conditions at the same point in time and whether model performance is being maintained. These are specialized to meet the needs of financial institutions intending to adopt PeopleFund's AI credit evaluation system.

The four XAI models developed and commercialized by PeopleFund this time are as follows: ▲ An XAI model that explains the positive and negative factors and their impact of each variable item on the borrower's credit evaluation results derived by the AI credit evaluation system using a mutual gate model, so that individual customers can understand them ▲ An XAI model that explains potential creditworthiness by analyzing the likelihood that a borrower who applied for a loan will become high-quality after several months ▲ An XAI model that explains the model's confidence level by evaluating in real-time how similar the data of potential borrowers currently entering for loan applications is to the data the AI credit evaluation system has learned in the past ▲ An XAI model that evaluates the degree of model performance degradation and the necessity of replacement according to changes in macroeconomic and financial market conditions.

Kang Min-seung, head of PeopleFund AI Research Center, said, "In a situation where social trust in AI transparency and fairness is still insufficient, if the algorithm's judgment results cannot be sufficiently explained, it is judged that AI credit evaluation technology will have difficulty expanding continuously, which led to the development of XAI models," adding, "We will focus on technological research to quickly detect and resolve newly discovered discrimination and unfair elements in the AI credit evaluation system."

This content was produced with the assistance of AI translation services.