Dementia is a global health challenge affecting millions, significantly diminishing the quality of life for people living with dementia (PLWD) and their caregivers. Reminiscence Therapy (RT) is a non-pharmacological intervention that enhances the well-being of PLWD by prompting them to recall and share past experiences through photographs, and artifacts. Although effective when personalized, scaling RT interventions remains challenging due to their labor and time-intensive design. To address this, we developed the Autobiography Gallery, an AI-powered system that uses Generative AI and Large Language Models to transform narrated memories into personalized photo albums. This thesis evaluates the effectiveness of AI-generated images in RT through feedback from dementia care stakeholders, identifying refinement needs, and proposing features like user-driven customization. We also provide design recommendations for scalable, AI-based RT interventions. Our findings highlight the potential of AI to enhance personalization and scalability in RT, advancing the development of accessible, user-centered solutions for dementia care.