the biggest AI change seems to be its ability to understand sentiment, like if someone is being negative it may have a harder time comprehending sarcasm, although I believe lab systems are basically at human levels at this point. The problem Alphabet have is the neural compute load for full AI is huge, so they can't fully understand everything, so they will probably seek to tag accounts as wrongthink and just ghost everything I'm guessing current moderation is just some word flagging, pattern analysis and account tagging I've had one account get auto deleted for normal speech, others seem to have fewer problems. other people can use words that I can't, I have to substitute "B" for black If they are auto-deleting you then just post positive shit for a few hours and it resets. It may be prudent to not expose every account to moderation, because I hit a limit of around 50 accounts I could make with one phone number