Maka Tsulukidze

The rapid proliferation of generative AI in higher education has outpaced evidence-based pedagogical frameworks. This paper presents findings from a structured faculty fellowship in which ten instructors across seven disciplines integrated AI into their courses and documented their practices. Fellows were selected through a competitive application process evaluated by an internal committee using a formal 100-point rubric. The program ran throughout the 2025-2026 academic year with monthly cohort meetings and a common nine-dimension teaching note template. Inductive cross-case analysis of seven completed Fall 2025 teaching notes, validated through iterative team review, identified four convergent pedagogical patterns: framing AI as a thinking partner rather than an answer generator; prioritizing productive friction over frictionless automation; treating prompt literacy as a core competency; and identifying verification as a persistent student weakness. These patterns informed a six-pillar Teaching with AI (TwAI) Framework and its companion A-V-E-R design pattern. The framework is offered as a practice-grounded contribution to AI-driven pedagogy and smart learning environments, with ongoing validation planned for Spring 2026.