"Faith" is a word that's tossed around a lot in religion. It's rarely defined, and when it is, the formal definition doesn't always match the colloquial one. First let's get straight what we're not talking about: "faith" can mean "trust" (as in "I have faith in you") or "belief system" (as in "the Christian faith"), and the similar-sounding "faithful" means "loyal" or "dependable" (as in "my faithful steed"). None of these definitions are relevant here. We're talking about faith as belief – the only question is what kind of belief.
It should be obvious that one should only believe a proposition X if there is sufficient evidence suggesting that X is true. I think it's helpful to think of belief in terms of a thermometer. The level of the liquid inside is the amount of evidence for a given position. Partway up the device is a notch marked "Rational Belief." If the evidence rises above that notch, the belief is rational. If it remains below that notch, the belief is irrational.
Based on this framework, here are two possible definitions of faith:
- Belief based on sufficient evidence
- Belief regardless of (and at times despite) evidence
The first of these is always justified, while the second is virtually never justified. (There are a few very basic exceptions. For example, I believe I'm not a brain in a vat, despite apparently having insufficient evidence to reach that conclusion. However, situations like this can often be resolved with Occam's razor: to believe I'm a brain in a vat would require unnecessary assumptions. In such cases, I would argue that Occam's razor constitutes its own unique form of evidence.)
What definition of faith do religious people adhere to? It depends on the believer, and I suspect that even within individuals the definition will subtly shift to whichever one is most advantageous in a given situation. If one asks about "blind faith," they'll flatly deny that the term applies to them. But when Christians say that "faith is a virtue," it's unlikely that they mean that "evidence-based belief is a virtue." When they use the term "unshakable faith," they probably don't mean "unshakable except in the face of sufficient evidence." And when Christians evangelize, they usually don't present evidence; instead they use the gospel message of sin and redemption as an appeal to emotion with faith at its core. Whether they realize it or not, they're advocating faith for its own sake, thereby encouraging people to believe irrationally.
Because faith is such a slippery word, I suggest dispensing with it as a generic term. There is "irrational faith," and there is "rational belief." There is no third category. Making this clear will force religious people to find reasons for believing. They may find flawed reasons, and they may be heavily biased in favor of their current views, but at least they'll be thinking. They'll be attempting to work their way up the meter to rational belief. And that's a start.