They don't.

To make sure that we're on the same page, make sure you understand what faith is (believing something without sufficient evidence, or contrary to existing evidence).

We use science, because it's demonstrably effective beyond all other human tools. Since the overwhelming evidence indicates that it works, it's not faith to rely on it.

Science is ultimately accurate by design. Let's say I were give you an inkblot test, and ask you "what do you see?" Most people would answer, depending on the pattern, something like "butterfly" or "face". That's actually a bad answer. A better, or objective answer is "I see a piece of paper that has blotches of ink on its surface, in a symmetric pattern due the ink spreading because of folding of the paper." Our minds understand things on familiar patterns and trends, which can often mislead us, or cause us to imbue what we're observing with our own narratives.

The point is - the human brain is incredibly susceptible to bias, inaccurate interpretation and error. Human memory is so unreliable that it ranks lowest, in terms of evidence quality, in courts of law. Things like anecdotes are almost entirely worthless as evidence.

The most important focus of science is to find ways to objectively verify, outside of the human mind, assertions and data - to reduce error and bias as much as possible. That way, whatever answers are generated have a relatively high accuracy.

Sure, other means of accumulating knowledge exist. We could consort with a Magic 8-ball for knowledge, however, the question remains, "how do we know it's accurate?". Science, as a methodology, is all about ensuring that we maximize accuracy.

Since skeptical atheists are intellectually honest, more often than not, and wish to ensure that their beliefs are actually true, they use science, because it works.