I also think these tools should credit the articles and products they rely on to produce the content so that users know where that information is coming from.
Yet they don't - ever wondered why? It is because likely, a lot of the content they use is not overly great quality, and so generic it is useless for all.
I tried Chat GPT to see how they would handle research articles. The paragraph noted one article when I specifically asked on the chat function. I could not find that article when I searched manually. Not sure where that article was pulled from.
And, the paragraph that came had errors throughout. I am sure the machine learning can improve the quality over time as the more we use the product the cleverer it becomes.