This paper explores the integration of Generative Artificial Intelligence (GenAI), particularly Large Language Models (LLMs), into educational settings to enhance educator productivity and student learning. While LLMs are increasingly used via conversational user interfaces (CUIs) for tasks like generating educational materials and providing feedback, this approach faces challenges such as the need for expertise in prompt engineering, ethical concerns, and limitations in handling complex tasks. To address these issues, the authors propose a framework for pedagogically sound and ethically responsible integration of GenAI into educational tools, emphasizing a human-centered design. They introduce Feedback Copilot, a novel tool that enables instructors to provide personalized qualitative feedback on students' assignments. The tool uses a set of assessment tasks, sample solutions, and graded student assignments, along with feedback quality standards, to generate customized feedback for each student. An evaluation involving 338 students demonstrated the tool's effectiveness in producing high-quality feedback. The study highlights the potential of GenAI-enhanced educational tools and encourages further exploration into user-centric GenAI applications. The framework includes application design and interaction design components, guiding the development of GenAI applications with user-centric interfaces. The research underscores the importance of educator oversight in feedback generation, particularly for lower-performing students, and emphasizes the need for ethical and pedagogically sound integration of GenAI in educational settings.This paper explores the integration of Generative Artificial Intelligence (GenAI), particularly Large Language Models (LLMs), into educational settings to enhance educator productivity and student learning. While LLMs are increasingly used via conversational user interfaces (CUIs) for tasks like generating educational materials and providing feedback, this approach faces challenges such as the need for expertise in prompt engineering, ethical concerns, and limitations in handling complex tasks. To address these issues, the authors propose a framework for pedagogically sound and ethically responsible integration of GenAI into educational tools, emphasizing a human-centered design. They introduce Feedback Copilot, a novel tool that enables instructors to provide personalized qualitative feedback on students' assignments. The tool uses a set of assessment tasks, sample solutions, and graded student assignments, along with feedback quality standards, to generate customized feedback for each student. An evaluation involving 338 students demonstrated the tool's effectiveness in producing high-quality feedback. The study highlights the potential of GenAI-enhanced educational tools and encourages further exploration into user-centric GenAI applications. The framework includes application design and interaction design components, guiding the development of GenAI applications with user-centric interfaces. The research underscores the importance of educator oversight in feedback generation, particularly for lower-performing students, and emphasizes the need for ethical and pedagogically sound integration of GenAI in educational settings.