This is a problem in communication: we struggle constantly to strike the proper balance between clarity and haste. When I tell you that I own a cat, I don't generally take the time to specify that it is of the domesticated Felis sylvestis catus variety; it's safe to assume that although you may be picturing a tabby when in fact I own a calico, you're unlikely to jump to the conclusion that I have a pet jaguar or mountain lion. That level of specificity is, under most circumstances, uncalled for.
But other entities are not so well-defined. What of deities? It's fair to say that both Hindus and Jews believe in gods, but do they both believe in the existence of gods in the same way that they both believe in the existence of grasshoppers? One might argue that their conceptions of the latter would be more consistent than their conceptions of the former. When one person says "god", it is often difficult to know precisely what he or she means. When discussing certain abstract concepts, some measure of increased clarity is called for.
And so to the current definitional problem. There are actually several disparate paradigms that can be used to disambiguate atheism from agnosticism. The way it's commonly understood, atheism is a belief that there is no god, theism is a belief that there exist one or more such things, and agnosticism is smack-dab in the middle.
This is incorrect.
I enjoy etymology, and I like my definitions to be etymologically correct. The word "atheist" comes from the Greek "atheos", literally "without god". Alternatively, if you just want to break it down the way the word is structured today, "theism" means "belief in the existence of a god or gods", with the prefix "a-" meaning "without". Now here's where most people get tripped up: properly understood, atheism isn't a belief the way theism is; an atheist does not believe in any gods, however he or she does not necessarily believe that those gods don't exist.
This is a sticky point, so allow me to clarify by analogy.
Imagine, if you will, that I've flipped a coin. The result is hidden from you, but you know that either the coin shows "heads" or it shows "tails". If I were to ask you, "Do you believe that the result is 'heads'?" what would you respond? What if I asked the same about "tails"? In this instance, many people might simply respond with, "I don't know." There is a problem here, however: that answer does not, in point of fact, address my question. I didn't ask about knowledge; I asked about belief.
In this situation, I would have to admit that I do not believe that the result was "heads". But—and this is very important—that does not mean that I believe that the result was "tails". In fact, I would also admit that I do not believe that the result was "tails", either.
Not believing that a statement is true does not mean believing that the same statement is false.
Atheism and agnosticism actually address different issues. Many people believe that there is a spectrum of belief that looks something like this:
ATHEISM ←→ AGNOSTICISM ←→ THEISM
The truth is that atheism and agnosticism do not fall on the same spectrum at all. Atheism deals with belief, as discussed above; agnosticism deals with knowledge. Let's go to back to our definitions.
The term "agnostic" was reportedly coined by T.H. Huxley ("Darwin's Bulldog") in 1869. It is also derived from the Greek, meaning "without knowledge". Put simply, knowledge and belief are interrelated, but it is possible to believe something without knowing, and it is even possible to know something without believing.
I am an agnostic atheist: I don't know that no gods exist, but I don't believe that any do. I'm also an anti-theist: I happen to believe that no gods exist—certainly no gods as defined by the traditional three omnis. But, my knowledge being imperfect, I cannot say that I know.
It is a straw man to declare that atheists claim (to either know or believe) that there is/are no god(s), and thus attempt to shift the burden of proof. An atheist is not making a claim: he or she is simply unconvinced.
Semper necessitas probandi incumbit ei qui agit.