I completely block the AWS address space on my servers because it's a major source of malicious probes and constantly siphons bandwidth at no apparent benefit to me. It looks like I'll have to consider less draconian measures now, if this is a source of useful messages and not merely a giant spam machine.
I'm just blocking originating connections from AWS on select servers. It's riskier in theory than in practice, as even the nonmalicious connections don't add much value. But you're right, I'm revisiting and will probably scale it back to just block web crawling, as no human activity seems to originate from that space.