On Math Overflow, someone asked for "A single paper everyone should read?"
and some answers were particularly nice to read for me, so I repeat it for you, ordered by how much math is needed (from none up to little):

• Paul Lockhart: "A Mathematician's Lament" shares my opinion about the math eduction disaster in schools. I think you should read this if you disliked your math classes in school or if you will ever have children (who will have to take a math class, then).
• Terry Tao: "What is good mathematics?" which is a short (10 pages) paper about the benefit we have from mathematicians different tastes and approaches. I recommend to every scientist reading the first 3 pages (the other 7 pages are only understandable with some background in mathematics).
• Freeman Dyson: "Birds and Frogs" which is a must-read for anyone interested in history and/or progress of mathematics.
• Misha Gromov: "Spaces and Questions" which is readable with almost no background, although might be funnier if you know basic differential geometry. It tells a dense story of geometric ideas and their development in history. And it doesn't take much time to read/skim it.
• Timothy Chow: A beginner's guide to forcing is a really gentle introduction to forcing.

Math Overflow is a new community website where mathematicians can discuss research problems. It is based on Stack Exchange, the software powering Stack Overflow, which does the same for computer science.