Not a linguist but you can still detect the unnatural way bots interact. If you respond somewhat abstractly and especially with sardonic humor for example, they seem to always take you seriously. They can't detect the subtleties of human intelligence.

Years before it seemed like actual humans (sock puppet operations overseas) but now seems like weird A.I. with some overseer human manipulation. Algorithms would conform to whatever social engineering goals. For example on Gab, take the natural impulse there of supporting whites and true history of WWII and such, but then exaggerating that to the point of absurdity (calls for violence) to try to justify attempts to give Gab a bad name and blur lines.

I mean, they have so much money to finance these operations. But they can't get enough soulless people to run them so are going to need more A.I,