Despite the plethora of anti-spam tools available to administrators running mail servers and the even larger suite of anti-spam software available for desktop PCs, spam isn't slowing down to any noticeable degree, and in fact some say that the problem is likely to only get worse. CS Professor John Aycock discusses this, and talks about how modern day spam filters are effective only because of the predictability of spam today. "Next-gen" spam might not be so.
I would be inclined to agree. Every once in a while I will notice a particularly effective spam message, that bypasses all the filters, and for a brief moment may look legitamte. It's hard for a PC to replace a human for such abstract things, which is why the problem will probably get worse before it gets better. It also talks about how modern spam bots can mine messages in order to become more effective. It's definitely worth a read.