Artificial Intelligence In Financial Decision Systems: Transforming Risk Assessment And Investment Practices In The Era Of Digital Scientific Culture

Abstract

Artificial Intelligence (AI) is quickly changing the way that financial decisions are made, especially when it comes to credit risk assessment and investment management. This paper suggests a powerful AI-enhanced framework that allows using ensemble-based machine learning models to make high-accuracy predictions of loan default risk. The Lending Club dataset, in the form of a pre-processed and balanced dataset, was used to test the models including Gradient Boosting, Random Forest, and Logistic Regression based on various classification metrics. The model that proved to be the most efficient was Gradient Boosting that had more than 92% accuracy and good precision-recall balance. The framework involves explainable AI (XAI) methods, such as SHAP and LIME, to guarantee transparency and accountability since both can provide both global and local interpretability of model predictions. The analysis of feature importance showed that such financial indicators as interest rate, amount of installments, and debt-to-income ratio may be considered as important factors of risk classification. The explainability tools incorporated in it serve to eliminate the ethical issues and further boost the confidence of the stakeholders in AI-driven finances. The findings of the study are that the integration of the predictive accuracy with interpretability would have a considerable impact on decision-making in high-stakes financial settings. Limitations like scope of dataset and fairness analysis are known, whereas future work will involve generalization of the models, adaptive learning and fairness auditing. This framework serves as a step toward more ethical, interpretable, and efficient financial technologies.