Ask HN: What is your workflow for filtering academic papers?

Given the large volume of AI papers circulating on ArXiv on a daily basis, how do you decide what is worth reading? Do you stick to a few known authors or only read what gets promoted on HN?