Problem is: We're soon on the second level where you are looking through papers that are based on AI slop nobody checked. How many layers deep can you reasonably go? Science is toast.
To echo: do not cite anything you have not at least skimmed and can verify it really exists and says what you say it says (or is about what you say it is about). Even if you collect citations from others' work (a legit practice), CHECK THEM.