Skip to content Skip to footer

Desperate ‘GoT’ fan programs A.I. to finish books for George R.R. Martin

So Game of Thrones is in a weird place.

With the final six-episode season reportedly not coming out until spring 2019 and no sign of any of the books from George R.R. Martin, Thrones fans have been left to their own devices to figure out what’s going to happen next.

Well, Zack Thoutt is one of those fans. He also happens to be a full-stack engineer and has programmed a recurrent neural network (RNN) to anticipate how the next book ends.

RNN’s are a type of machine modeled after the human brain that can take inputted data and react accordingly.

Thoutt explained his RNN to Motherboard,

“With a vanilla neural network you take a set of input data, pass it through the network, and get a set of outputs. In order to train these models you need to know what the model should ideally output, which is often called your labels or target variables. The neural network compares the data it outputs with the targets and updates the network learns to better mimic the targets.”

Ok, so what does this actually mean?

Thoutt plugged the Game of Thrones books into his RNN and created his own sequel to George R.R. Martin’s series. Luckily for GRRM the technology isn’t perfect, at times it references characters that have already died. The engineer explained the problems of the RNN,

“It is trying to write a new book. 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. The model is striving to be a new book and to take everything into account, 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.”

Despite some grammatical and narrative flaws the RNN is still pretty damn effective. It’s also worth a couple laughs, the first sentence it churned out would reveal a pretty large twist (possible spoilers and fan theories ahead),

“‘I feared Master Sansa, Ser,’ Ser Jaime reminded her. ‘She Baratheon is one of the crossing. The second sons of your onion concubine.'”

Sansa is a Baratheon! But more importantly, what or who is this onion concubine? I’m intrigued.

Even more wildly, the RNN followed through on a big fan theory that Jaime may turn on his sister, as Cersei’s prophecy says one of her brothers will kill her,

“Jaime killed Cersei and was cold and full of words, and Jon thought he was the wolf now, and white harbor…”

Thoutt told Motherboard this was purely coincidence,

“I guess that validates that anything can happen in Game of Thrones. I didn’t feed it anything from fan theory websites, only the books.”

GRRM better hurry the fuck up and get to work. As entertaining as season 6 of Thrones was, there were definitely a clear departure from Martin’s pacing and dialogue.

It’s becoming pretty doubtful that Martin will ever actually finish these books.

In the meantime, maybe we should all build our own recurrent neural networks to see what happens.