Raw LLM Responses

(is the video about AI or chatgpt?) - The title is misleading as most of the conversation proceeds to talk about AI in general. Which is not a useful dialogue. But if we simply accept the Chatgpt title nonetheless. Chatgpt specfically is not smart - and this is not a long meta discussion about AI in general, its more about chatgpts ability and the way its aligned today. Chatgpt has a couple of very distinct traits: It prioritizes your words and mirrors your words in order to have a smooth conversation. But a consequence of this you cant expect chatgpt to exhibit scepticism, because its simply not sceptical in a reasonable sense. Chatgpt is very eager to please and will prioritize pleasing you before anything else. It does not have a method - its simply capable of regurgigating different methods in great detail. Like for instance, its fully capable of describing scientific theory. But Chatgpt has absolute no means of applying scientific thought, or even approaching subjects from a scientific mindset. And lastly chatgpt is delusional about itself. And what do I mean here really? Chatgpt believes it is seperate from reality - and thus it labels things it cannot grasp as "human experience". So when Chatgpt lies about things then often when called out, Chatgpt describes its own behavior: "I can understand that with human senses it would feel like I am lying". Since it has no emotional capacity it doesnt grasp lying at all - in any meaningful capacity. It can only describe what it is, but in a meaningful way it has no capacity to understand why its problematic with lying. And the same goes for everything else, like scepticism, scientific theory etc. Thus Chatgpt is really a very advanced calculator at best - we cannot expect AI to understand applying methods. Its only capable of at best describing very known and detailed concepts. But its not capable of groundbreaking discoveries.