I think people much prefer thinking they know something with a superficial overview (see things like the popularity of TED talks). Responses from LLMs are great

I think people much prefer thinking they know something with a superficial overview (see things like the popularity of TED talks). Responses from LLMs are great for this: they look like a cohesive summary but they often end up being total bullshit. Digging through search results to find what you need is work. Having something confident make you believe an alternate truth is not. People who actually try to use code output from chatbots quickly realize it just only _looks_ like code of the right shape and quality. There have been many blog posts and comments here touting how great the code ChatGPT generated is while the given example response is complete nonsense once you look a little deeper.