Imagine learning online, but the tutor doesn't speak your language – not in the literal sense, but in terms of understanding your unique needs. This is the reality for many d/Deaf and Hard-of-Hearing (DHH) students. While intelligent tutoring systems (ITS) hold immense potential, they often fall short of addressing the communication and cultural nuances of the DHH community. However, a recent study explores the potential of Large Language Models (LLMs), like the technology behind ChatGPT, to power AI tutors that truly connect with DHH learners. Researchers experimented with giving AI tutors different 'personas,' including backgrounds in DHH education. The results were fascinating: DHH students found tutors with DHH-related experience significantly more trustworthy and human-like. Why? Because these AI tutors asked relevant questions, offered personalized feedback, and demonstrated an understanding of DHH culture. This made students feel seen, heard, and more engaged in the learning process. However, the study also reveals challenges. Current LLMs struggle with the multimodal nature of sign language. Text-based descriptions of signs simply can't replace the richness and expressiveness of visual communication. Moreover, the lack of detailed information about the AI tutors' background raised trust issues. Students wanted more transparency regarding the AI's experience within the DHH community, including their ASL proficiency. This study underscores both the potential and the limitations of AI in education. It offers a glimpse into a future where AI tutors are not just instructors, but personalized learning companions who bridge communication gaps and empower DHH learners. However, to truly unlock this potential, we must address the need for multimodal support, transparency, and cultural sensitivity in designing AI-powered educational tools.