This study aims to forecast future sequences of inflation and the investor sentiment index using artificial intelligence in Iran and Iraq. In terms of purpose, the research is applied, and in terms of nature and method, it is descriptive-analytical, conducted through a hybrid approach combining artificial intelligence and econometrics. The statistical population comprises monthly data on inflation and the investor sentiment index in Iran and Iraq, with the statistical sample including all available observations over the 10-year period ending in October 2025. Data were collected using a documentary method from official statistical sources and reputable databases. In the first stage, a Long Short-Term Memory (LSTM) recurrent neural network was employed to forecast future sequences of both variables; 80% of the data were allocated for training and 20% for testing, with forecasts evaluated across seven future sequences. The results indicated that the model demonstrated more stable and reliable performance in forecasting the investor sentiment index in Iran, whereas in Iraq, prediction errors gradually increased as the forecast horizon extended. Furthermore, the findings from the Error Correction Model revealed that a long-run equilibrium relationship exists between the variables in both countries, and the system returns to its equilibrium path at an appropriate speed. Overall, the findings suggest that integrating artificial intelligence and econometric methods can provide a suitable tool for inflation forecasting, with results applicable to economic policymakers and capital market participants.