I'm sure most people on Nostr have already encountered the airdrop spammers. Their spam is very unsophisticated, the note content hash is identical in a lot of cases and could be screened out just by this.

Partial matching would be the next followup countermeasure.

AI matching would be another, but I think probably at first it would be unwise to mute but instead have it just put a ? flag on suspected posts and a confidence rating, and users could then train the AI progressively to become smarter.

Something that's really important for people to understand. AI cannot be creative, because it is not alive and has nothing at stake. Creativity is forced by survival, because it tends to be "out of left field" and "off the wall". It is things that come out of places that seemed extremely unlikely to yield anything of value.

This is something that an AI cannot do, because it doesn't have enough general intelligence to do it. All the AI compute in the world can't replace the devil at your back.