This paper examines the role of enchantment in the design of AI things by constructing a taxonomy of design approaches that increase or decrease the perception of magic and enchantment. The authors start from the design discourse around recent AI developments, highlighting interaction qualities like algorithmic uncertainties and errors, and their relation to the rhetoric of magic and supernatural thinking. By analyzing 52 student design projects from two editions of a Master's course in design and AI, they identify seven design principles and unpack their effects in terms of enchantment and disenchantment. The paper concludes by articulating how this taxonomy can be used by design/HCI practitioners to support exploration and reflexivity. The paper discusses magical thinking in the design of technology, emphasizing how AI products are inherently enchanting due to their complexity and opacity. It explores two distinct approaches to design: one that enhances enchantment through magic metaphors and stage magic principles, and another that reduces enchantment by manifesting mechanisms and materializing beliefs. The authors argue that while magical thinking can be used to create engaging and seamless interactions, it also carries risks, such as deception and the obscuring of AI's actual capabilities. They propose a taxonomy of design principles that can help designers navigate these dynamics, balancing the potential of enchantment with the need for transparency and critical reflection. The taxonomy is developed through a conceptual-to-empirical approach, using 52 student projects as a corpus. The authors identify categories such as applying magic metaphors, applying stage magic principles, manifesting mechanisms, and materializing beliefs. They also introduce additional categories like summoning AI as a supernatural entity and playing with AI. The taxonomy is presented as a framework for understanding and designing AI products, emphasizing the importance of balancing enchantment with critical awareness. The paper highlights the need for designers to be honest about AI's capabilities and to avoid deception, while also recognizing the value of magical thinking in creating engaging user experiences. The authors conclude that the taxonomy provides a valuable resource for design practitioners, educators, and researchers to inform future designs and practice reflexivity.This paper examines the role of enchantment in the design of AI things by constructing a taxonomy of design approaches that increase or decrease the perception of magic and enchantment. The authors start from the design discourse around recent AI developments, highlighting interaction qualities like algorithmic uncertainties and errors, and their relation to the rhetoric of magic and supernatural thinking. By analyzing 52 student design projects from two editions of a Master's course in design and AI, they identify seven design principles and unpack their effects in terms of enchantment and disenchantment. The paper concludes by articulating how this taxonomy can be used by design/HCI practitioners to support exploration and reflexivity. The paper discusses magical thinking in the design of technology, emphasizing how AI products are inherently enchanting due to their complexity and opacity. It explores two distinct approaches to design: one that enhances enchantment through magic metaphors and stage magic principles, and another that reduces enchantment by manifesting mechanisms and materializing beliefs. The authors argue that while magical thinking can be used to create engaging and seamless interactions, it also carries risks, such as deception and the obscuring of AI's actual capabilities. They propose a taxonomy of design principles that can help designers navigate these dynamics, balancing the potential of enchantment with the need for transparency and critical reflection. The taxonomy is developed through a conceptual-to-empirical approach, using 52 student projects as a corpus. The authors identify categories such as applying magic metaphors, applying stage magic principles, manifesting mechanisms, and materializing beliefs. They also introduce additional categories like summoning AI as a supernatural entity and playing with AI. The taxonomy is presented as a framework for understanding and designing AI products, emphasizing the importance of balancing enchantment with critical awareness. The paper highlights the need for designers to be honest about AI's capabilities and to avoid deception, while also recognizing the value of magical thinking in creating engaging user experiences. The authors conclude that the taxonomy provides a valuable resource for design practitioners, educators, and researchers to inform future designs and practice reflexivity.