The trouble in my view is that the only way to know that the answers you're getting are accurate and not misleading is to study up on the answers elsewhere - which is a great habit to nurture, but is also precisely why these tools tend toward uselessness in their "general AI" bids. If I can't know how the answer was built, or how good that answer is, there's no point asking it - I'll just do my own reading and apply appropriate discernment as I go.
To be fair, hardly anyone does this today, nor did they before LLM-based chat bots... So it's a moot point, because society is largely doomed anyway. But a moot point can still be a valid one.
I also think the author makes a good point that we frequently confuse performance for competence. "It does a really good job at <X>!... or at least does a damn fine job of mimicking someone who acts like they do a really good job at <X>!"
By way of analogy, consider Elon Musk - by all appearances, he's a genius and is saving humanity - but by dint of his narcissism and largely smooth-brained approach to... well... everything... he's running all of us into an earlier planet-size grave than is necessary. His performance is fantastic, his competence is nonexistent.