Lung cancer is one of the most common types of cancer among various malignancies. It is very hard to detect lung cancer through simple assistive tools, which has been observed in recent studies. Hence, this disease still requires an effective and efficient diagnostic tool for providing better health care for mankind. This work presents an advanced lung cancer detection method using ensemble learning and explainable artificial intelligence with Gradient-Weighted Class Activation Mapping visualization for histopathological microscopic images. The adopted method utilizes three powerful state-of-the-art models, including InceptionV3, ResNet50, and VGG16. The transfer learning of these models extracts key features from microscopic images, allowing for efficient and accurate classification. The Gradient-Weighted Class Activation Mapping integration offers a concise visual representation of the model's prediction outcomes. The outcome of this detection highlights the region of interest areas for better understanding and clarity, which may assist the clinical experts in making decisions. The publicly available LC25000 lung cancer dataset has been accessed for the experiment. The outcomes of the proposed approach demonstrate an enhanced accuracy of 99%. The method's better performance has also been validated by means of other assessment criteria including recall, precision, and F1 scores.