Extant chatbots such as ChatGPT and Bard are currently able to converse with humans in natural language, demonstrating impressive linguistic responses. Or so it seems. I critically examine artificial intelligence systems such as these chatbots through examples of dialogue. When taking a systems view of AI, there is a vast and unique human culture in the environmental surroundings of the AI "system" (its "negasystem") that is not accessible to extant AI "systems." These generative AI "systems," based on large language models, are trained with trillions of signs created by humans in the form of digital text and images as part of their machine learning from which they construct their unique neural networks. However, AI "systems" do not understand well, if at all, the meanings of those signs that we associate with our human experience of the world and our culture (i.e., in the AI negasystem). Similarly, we humans do not understand well the inner workings of an AI "system" (its neural network). Teachers and students in education must be very careful and cautious when using such AI "systems." Are we dupes? Or not? Without thinking critically and checking facts independently, we can be fooled by responses of those AI "systems."