I assumed you guys knew what I am talking about. Here is one of many sources on Logical Fallacies I have.Read and it will soon become obvious what I am talking about. http://www.nizkor.or...ures/fallacies/Whoever coded shadowhawk 0.9 beta is quite an impressive programmer, however I doubt it could come anywhere near passing a turing test. I'd suggest integrating public facebook databank correspondence so that it becomes less repetitive.
They could also get it to use terms like "logical fallacy" only where it is refering to a putatatively logical, but wrong, argument. The mechanical reiteration of it's three stock responses, often one after the other and without any attempt at logical analysis does tend to give the game away.
Where is the evidence? None
Once again it is just quoting someone else. It's library of stock responses is really quite limited. This is just a fairly simple summary of the ordinary definition of logical fallacies as given to first term first year students, but it doesn't justify the use this low grade AI is making of the term. It doesn't seem to be able to distinguish a comment from an argument. Perhaps it just assumes all posts are arguments.