
The introductory chapter asks whether computers can create new, original stories. Basic methods include chopping and reassembling previous stories, following flow charts of plot elements, and coding rules to generate story outlines, actions and dialogue. By the start of the 1990s, three books had been published that were claimed to have been written by computer programs. For the next two decades, a few researchers in computers and creativity coded programs that would work like human authors to plan, draft and revise well-formed stories. Then, in 2019, the OpenAI company announced a neural network program, GPT-2, that can continue the style of any writer without the need for expert programming. However, neural network text generators have fundamental weaknesses: they have no common-sense knowledge of the world, they cannot reflect on what they write, and they cannot explain their inner workings. The book explores a grand challenge to design successful story machines that write like human authors.