Advances in conversational AI have improved human-tech interaction by enabling coherent, context-aware responses. However, maintaining contextual awareness in longer conversations remains challenging. This study presents a conversational AI model based on an enhanced Seq2Seq architecture with attention mechanisms and bidirectional encoding, trained on the Cornell Movie Dialogues dataset. The model achieves 80% accuracy, significantly outperforming traditional Seq2Seq models (50–60%) and attention-based models (65–75%). Utilizing a Global Luong Attention Mechanism, it exceeds the performance of models with basic attention (70–75%). This work aims to enhance applications in virtual assistants, customer service, and interactive entertainment while addressing class imbalance and preserving context. Future research will focus on optimizing the model for real-world applications and improving dialogue quality.