The less said, the easier it is for a language model to approximate a useful thread comment for the purposes of mass propaganda. “These graphics look terrible.

The less said, the easier it is for a language model to approximate a useful thread comment for the purposes of mass propaganda. “These graphics look terrible. I will never play this game.” “$Candidate is a corporate shill and everyone knows it.” “I can’t wait for $Artist’s next album! They’re sooooo good!” Doesn’t need to be an extensive, well thought out comment to drive thought and discourse. GPT2 is good enough for that.