The wait for "The Winds of Winter" continues, as enthusiasts are hopeful that the highly anticipated novel will be released sooner rather than later. Recently, a software engineer was tired of waiting for the next installment, so he trained a recurrent neural network to finish the sixth book instead.
Fans of George R.R. Martin have been looking forward to the sixth volume of "A Song of Ice and Fire" series. While it is still coming, the author has pushed back the release date a number of times and instead released a few secondary books in the meantime. It is certainly no one's business if he chose to publish other novels, but it is understandable why readers are getting a little disappointed and antsy.
Martin started penning down his epic fantasy in 1991, and "A Game of Thrones" was published five years later. Initially intended as a trilogy, he has so far delivered five out of a planned seven volumes. The most recent one, "A Dance with Dragons," took five years to compose and was then released in 2011. Six years later, everyone anxiously waits for the cliffhanger resolutions, new battles, and the actual coming of the winter.
It remains unclear when "The Winds of Winter" release date will be. Martin stays tight-lipped about the next installment. Truly desperate readers, however, can grab a copy of the sixth novel from programmer Zack Thoutt.
Having just completed his course on artificial intelligence and deep learning, Thoutt wrote on his blog that he needed a "new project to keep" his skills sharp. He further stated, "One episode into season seven of 'Game of Thrones' (I'm a huge fan!), the idea hit me — why not train a network to write new chapters for the book we're all waiting for?"
The software engineer had been watching the "Game of Thrones" series while studying recurrent neural networks, a type of artificial intelligence that consists of algorithms learning from the input of date they have provided with. To address the ongoing demand for "The Winds of Winter," he decided to conduct a little experiment.
With the 5,376 pages of Martin's previous books from the series, his AI produced the next chapter. Apparently, it did not match up the author's style, but its plot was beyond impressive. Even the sentences that didn't make sense contained a fascinating poetic aspect to them.
Thoutt continued to explain on his blog, "My GoT AI gave us a small taste of what is possible with deep learning. Every day researchers around the world are coming up with new network architectures to better address problems that we face in society. The possibilities of AI going forward are vast, and I'm excited to be on the cutting edge of the technology behind the AI movement."
