Doesn't this hugely depend on what you are doing at the moment?
- If you have a well-defined project, you have to read a small number of papers very, very thoroughly
- If you just want to keep up with the field, you might just take a few notes of a paper, and skip the technical details
- If you just want ideas/brainstorm/think of applications of a method, it might be enough to read abstract, introduction, and conclusion (or just the figures). That's enough for me to keep it in the distant memory, and come back to it if necessary.
Edit: This might think a bit too confident...
I'm currently a little bit daunted by the amount of stuff out there that would be worth reading, and it's almost paralysing me a little.
That's probably a familiar feeling for many (most?) of us.