I've just been reading about some guys at MIT who've written a program that will automatically write computer science papers.
It's quite clever, as it draws graphs and all sorts of things :) Anyways... it's in the news because its paper on Rooter: A Methodology for the Typical Unification of Access Points and Redundancy was accepted to a conference! Unfortunatly, due to all the news coverage the conference people have realised they look a tad foolish and have revoked their invitation to speak. One might think that it was a bit silly to announce it was a fake paper publically before the conference. Although I suppose they had to to gt people to donate money to them so they could go and present it!
You can read all about it on the project's website.