Given the quality of a lot of Facebook comments any of the Loebner Prize bots would have a reasonable chance of passing as human. (http://www.loebner.net/Prize

Given the quality of a lot of Facebook comments any of the Loebner Prize bots would have a reasonable chance of passing as human. (http://www.loebner.net/Prizef/loebner-prize.html http://www.loebner.net/Prizef/loebner-prize.html) A bit of tweaking - some learning and some Markov chains - and you'd have something reasonably solid for that narrow gap. Obviously there are trivially easy ways to trip up bots, if you suspect they are bots.