As AI systems become increasingly autonomous, ensuring their trustworthiness is critical. We propose a hybrid human-AI approach to decision-making that leverages both human and machine intelligence to achieve high accuracy while maintaining transparency and accountability. Our approach uses machine learning to provide decision recommendations to humans but also explains the reasons and uncertainties behind recommendations to enable human oversight. Humans can approve, reject or edit recommendations based on this information and their own judgment. We evaluate our method on sensitive decision tasks like financial loan approvals and medical diagnoses. Results show our hybrid approach outperforms either human or AI alone in accuracy and user trust, demonstrating the promise of hybrid models for responsible decision automation.