I follow the same approach as newapollo. The spambots are usually easy to spot as they contain a spurious link and the account has normally been created on the day of the post. These are normally deleted quickly by the moderators when flagged.
A bit harder to differentiate are the posts which actually answer the topic question (to varying degrees of success). Sometimes I get the impression it is a real person posting who has used AI to come up with an answer and just pasted it as a reply. Other times it appears as if it is a fully automated (and reasonably accurate) reply. In either of those cases I don't flag it if it does not affect the topic.
Maybe VM is trialling some AI bot responders on here!