Public responses to artificial intelligence (AI) are shaped not only by individual characteristics but also by the specific type of AI under consideration. Yet most existing research treats AI as a unified construct, obscuring meaningful variation across distinct AI categories. The present study examined attitudes (Acceptance, Fear) and apprehension (Implied Malice, Undesirability, Unpredictability) toward four theoretically distinct AI types (General AI, Large Language Models, Personal AI, and Institutional AI) in a within-subjects design with 555 adults. Repeated-measures ANOVAs revealed that AI type significantly shaped both attitudinal and apprehension profiles. Personal AI was associated with the highest acceptance and the lowest apprehension across all three dimensions, while General AI evoked the highest fear and apprehension. Large Language Models showed a pattern of attitudinal ambivalence, with Acceptance and Fear closely aligned in magnitude. Institutional AI was characterized by elevated Implied Malice and Unpredictability comparable to General AI. Across all AI types, Acceptance was negatively and Fear was positively associated with all apprehension dimensions, with moderate effect sizes confirming that apprehension is related to, yet distinct from, established attitudinal constructs. The internal coupling among apprehension dimensions was significantly stronger for Institutional AI, suggesting that apprehension is not a fixed psychological structure but one that is dynamically reorganised by context. These findings underscore the importance of treating AI as a heterogeneous class of technologies and provide a basis for context-sensitive governance and communication of AI systems.