Teachers still want practical, instructional answers about generative AI, according to research capturing educators’ classroom-centered questions. In fall 2024 discussions facilitated by EdSurge researchers, 17 teachers from around the world shared how they were or were not using generative AI in instruction and what they needed for it to improve learning outcomes. The study highlights “air of indifference” as a dominant sentiment, with educators asking a basic implementation question: what can AI do to help them teach specific subjects more effectively? A fourth-grade math teacher’s concern about how AI could support elementary students’ math learning illustrates the recurring tension between marketing hype and classroom utility. The research also positions teacher evaluation as expert practice—teachers are not simply rejecting innovation, but testing whether a tool solves a concrete instructional problem. That creates a roadmap for institutions: AI policy and training must connect to subject-specific pedagogy, not just generic guidance about AI safety. For higher education stakeholders supporting K-12 partnerships and teacher preparation, the key development is that adoption depends on instructional clarity, especially when teachers fear misuse or incomplete understanding of student AI use.