Those desperate to find out what happens next in the world of Westeros would usually turn to the book series to find the fate of their favourite characters.
However, the HBO show has overtaken the storytelling of George RR Martin’s novels and fans are left without their Game of Thrones fix for a year for the TV show, and until 2019 for the book.
Thankfully though, artificial intelligence works a lot quicker than George RR Martin, and following the conclusion of season seven, a bot has analysed all of the previous texts from the author to predict and write the 2019 instalment; Winds of Winter.
Software engineer Zack Thoutt created a type of AI known as a recurrent neural network which was fed the 5,000 or so pages of Mr Martin’s previous books, and this is what it believes happens next.
Warning: potential spoilers are ahead.
The bot believes that some fan theories do play out.
As seen in the last episode, Jaime leaves Kings Landing, presumably to go fight up north, but the book believes that he will return to kill Cersei.
Chapter five of the ‘new book’ begins: “Jaime killed Cersei and was cold and full of words, and Jon thought he was the wolf now, and white harbor…”.
The AI bot also believes that Jon Snow will ride a dragon, and perhaps most shockingly, Varys kills off Danaerys with a dose of poison.
While the storyline follows largely from Mr Martin’s previous works, the AI bot does throw in a few unexpected surprises.
For example, it predicts that Sansa Stark turns out to be a Baratheon.
Written from the perspective of Tyrion, chapter one states: “I feared Master Sansa, Ser,” Ser Jaime reminded her. “She Baratheon is one of the crossing. The second sons of your onion concubine.”
While it is impressive that an AI bot can muster up a novel, the story can often be too far-fetched, even for Game of Thrones, such as the long dead Ned Stark returning without explanation.
The sentences too are often clumsy and make little sense.
One sentence reads: “It made Ned better stop until the fire was falling, standing beneath the arch of a shattered still distant field where the shadow tower paid the camp behind.”
Zack Thoutt said that a lot of work would need to be done on the neural network to leave authors out of a job.
Mr Thoutt told Motherboard: “It’s obviously not perfect. It isn’t building a long-term story and the grammar isn’t perfect.
“But the network is able to learn the basics of the English language and structure of George RR Martin’s style on its own.
“A perfect model would take everything that has happened in the books into account and not write about characters being alive when they died two books ago.
“The reality, though, is that the model isn’t good enough to do that. If the model were that good authors might be in trouble … but it makes a lot of mistakes because the technology to train a perfect text generator that can remember complex plots over millions of words doesn’t exist yet.”