An ML model for millet yield prediction under different cropping systems shows an R² = 0.91 and RMSE = 0.25 t/ha. If observed yields are 2.8, 3.1, and 3.4 t/ha, and predicted are 2.9, 3.0, and 3.3 t/ha respectively, compute the Mean Absolute Error (MAE) and explain the model reliability under climate variability.