Analysis of the capabilities of predictive artificial intelligence models in corporate risk management

This study examines the use of predictive artificial intelligence models in corporate risk management. Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) architectures are evaluated in forecasting five widely used market risk indicators based on daily financial time series for Apple Inc. Model performance is assessed using a unified experimental framework and out-of-sample data, with accuracy measured by Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE). Results show that predictive effectiveness varies across risk measures and model types, and no single model consistently outperforms the others. These findings highlight the need to align model choice with the structural properties of the specific market risk indicators being predicted.