We had a chatbot on here recently - among other places, on the Syria thread. It was fairly easy to spot because it uttered nothing but platitudes. Yeah yeah I know. Seriously, though, it was just picking up on vocab and general attitudes and saying things like (I paraphrase) 'All interested parties need to come together to solve the problem.' Was it anthony243? (Hard to search for posts by banned users.) Anyway, it was an old gen chatbot. Not remotely interesting though some people were polite to it.
Then came Tay.
Microsoft’s racist chatbot returns with drug-smoking Twitter meltdown
Not here, not yet.
And then along came
yooper. I confess for a while I was convinced he was a new gen chatbot and fascinated by the idea. If I were designing a chatbot I'd factor in crap like bad punctuation and occasional misspelling to lend an air of verisimilitude. I'd also give him a bit of back story. Yooper's bloody-mindedness, including repeatedly posting something critical of the Trump Wall as if it were an endorsement, liking posts that insulted him, and picking up on odd words to post something irrelevant, all seemed to point to a heuristic bot. It wasn't until krtek a houby identified him as a long time serial offender on other forums that I was convinced he wasn't a bot, though by then he'd started to write some mildly creative posts for a change.
So it occurs to me: how can you tell if you're dealing with a bot?
