In a way that is what we humans do way too often. People talking about stuff thinking they know, but not actually knowing a lot. I guess that is where it learned that from.
But, after you've been exposed to this a few times, you start to get a sense for it. Most people have a "bullshit meter" which gets calibrated over time, and it's remarkably good at sniffing out the people who are out of their depth.
ChatGPT is somewhat different in that its level of command of the English language doesn't change in response to its level of knowledge about the topic it's asked about. It most likely has the most thorough command of the English language of any entity ever born or created; that is exactly and specifically what LLMs are supposed to be good at. Because of that, it can speak "perfectly" no matter the topic; it can be prompted to explain things in "simple terms", but remain eloquent. It doesn't fall back on jargon to mask its ignorance, and it doesn't breeze past important concepts without bothering to explain them. As a result, it communicates exactly like an actual domain expert would - carefully, precisely, and simply. This is wonderful for clear, effective communication, but it completely subverts the bullshit meter.
In some ways, maybe this could be a blessing. Media and marketing organizations have gotten very, very good at sitting in the "happy language zone" as they straight-up lie to our faces. Perhaps just flooding the zone with language which doesn't trip the bullshit meter and still manages to be empirically wrong could cause people to readjust and redevelop the heuristics they use to evaluate the trustworthiness of information.