As we know, we already have computer programmes that can play chess so well that professional chess players have been convinced that they’re playing a human. But that is different from writing a book, isn’t it?
In chess the pieces are bound by the number of moves they can legally make. The pieces have different values, with the pawn being the least valuable and the King being the prize. Once a piece has moved, there’s a finite range of possible consequences that result from that move. Finally, there is a finite (if very large) number of possible moves that can be played in any game. Once those variables have been fed in, the difference between the computer and the human is that the computer can calculate all possible moves, and the possible consequences and responses to them, far more quickly than the person.
Advances in artificial intelligence (AI) mean that computers are able to take on more and more interpretational tasks. That means the computer doesn’t just analyse data to find differences, but it also interprets what the data means, within the parameters set by the programmer. In other words, it is mimicking human interpretation. If it can work out how we understand things, then the next logical step would be to generate data that we understand. A story, basically, is data that we understand.
If that was as far as things went, we authors wouldn’t have much to worry about, but it isn’t.
OK, it isn’t War and Peace, but it’s a start. I’ve read children’s stories with less meaning. From its analysis of the data, Scheherazade “understands” that those events must take place in that order.
What is missing from that little description is motivation. Why did John enter the room? Why did he sit down? We understand that all actions have some motivation that makes them occur. Can you teach a computer that? Does the computer know why it is playing chess? No, but that doesn’t mean that a computer can’t be taught that there are a number of possible motivations for murder, to pick out just one popular genre in books.
Murder can be motivated by greed, jealousy, desire for revenge, fear of discovery of one’s crimes etc. These can be taught to a computer. Soon the computer could structure a story along the lines of “John entered the room where the money was kept. John murdered the man and stole the money”. Still not a great novel, but heading in the right (or maybe wrong) direction.
Now, I can’t speak for all authors, but more than one of my books has been triggered by a “what if” question. "The Charity Thieves" asks what if the wrong sort of people could get inside the fundraising arm of a charity? The rest of the book is spent answering that question (in an entertaining and adventurous story).
Not only does Whim ask the questions, it then assesses the answers to decide which scenario would be most appealing to an audience. In other words, it not only generates ideas that might be considered creative, but it also assesses their viability. Which means it is self-critical. Given some of the dross I’ve read over the last few years, some authors could do with learning that ability!
But of course, books aren’t just a series of random ideas. They are filled with metaphors, humour, tragedy, sarcasm and a whole load of other literary devices that make them interesting. You can’t teach a computer that!
The Metaphor Magnet, by the way, has public access, so if you click on a link within the article you can open a page that allows you to input a phrase and it will generate metaphors based on it. Have a go - but not until after you’ve finished reading the rest of the blog. It’s quite addictive and I don’t want to lose you just yet.
What about the other devices, such as sarcasm? They're way ahead of you. Oren Tsur, Dmitri Davidov and Ari Rappaport of Jerusalem University have created the catchily named Semi-Supervised Recognition of Sarcastic Sentences, or SASI for short. At the moment SASI is only capable of recognising sarcasm, but that is the essential precursor to using it in sentences.
Ah, but what about emotions? Computers don’t “feel” the way people do. so they can't describe feelings.
The whole thing hinges on understanding how the brain processes information, then translating that into computer algorithms that do the same. This was how we got started with computing in the first place, with Charles Babbage understanding how humans did arithmetic calculations then building a computer that did the same. But of course, we’ve progressed a long way since 1822, when Charles Babbage built his first “difference engine”.
Computer scientists, working with neuroscientists, are getting better and better at understanding how the brain works, and hence developing the necessary algorithms. So, how long before they understand the creative process that an author goes through to write a novel?
Of course, understanding how it is done isn’t the same as actually doing it. After all, plenty of people know how to tell jokes but very few have the talent to become professional comedians. Plenty of people make up stories, but that doesn’t mean that they can write a book; at least not a good book, that someone will actually pay money to read (like mine).
Seriously, I doubt that the age of the computer written novel is imminent, but I’m pretty sure that the big publishing houses are already licking their lips in anticipation. Imagine, not having to share their sales income with us pesky authors, demanding to be paid for the weeks, months and sometimes years of hard work we put into creating our books. So much more money for the shareholders. If the computer programs are good enough, they won’t even need to employ editors and proof readers. That’s even more profit! As for Amazon, Kobo, Smashwords etc - cut out the middle man (or woman) that is the publisher and go straight to the computer!
Even worse than that, people may not even care that a book has been written by a computer. If the computer can consistently produce work of quality, then it would be worth paying for. After all, we don’t care that our cars are built mainly by robots these days.
I’m not saying that it will happen this year or this decade, but I would say it might happen within the next 20 years.
Every time we use the internet to purchase something we are using a tool that has replaced human sales staff. The order is fed into the automated warehouse where a machine can pick it from the shelf and deliver it to the packing point. A machine can pack an item just as well as any human and then deliver it to a lorry, which can be programmed to drive itself to a distribution centre, where the bar code on the package can be read and it will be moved on in another driver-less lorry or delivery van. Or maybe it will be delivered by drone, which is something that Amazon is already working on.
All of that is feasible TODAY. Only the law needs changing to allow the driver-less vehicles onto the road and the "human free" supply chain can become a reality. Just this week the British government announced an £8.1 million trial of driver-less lorries on our motorways, which will start in 2018. If I were a lorry driver I'd start to worry about the future of my job.
Every time a fresh innovation is made, someone confidently predicts that whatever it is will never replace whatever it is designed to replace. But every time they make this prediction they are soon proven wrong.
The train replaced the horse and cart, as did the car. The lorry replaced the narrow boat and barge as a primary cargo carrier. The aeroplane replaced the passenger ship. The cinema replaced music hall and was, itself, threatened by TV. CDs replaced vinyl records and music downloading and streaming is replacing CDs. New cars don’t even have CD players installed because the manufacturers no longer think they are needed.
DVDs replaced the video tape player and have themselves been replaced, once again, by downloading and streaming. The e-book hasn’t yet replaced the paperback, but each year more people switch. The internet and cloud computing are replacing the physical library as the primary source of information. The march of technology is inevitable, only the speed of change is in question.
This may all seem very dystopian and there have been numerous books written and films made about the rise of computers and robots and the potential consequences of that rise, but just because it is dystopian it doesn’t mean that it can’t happen.
On 8th June 1949 George Orwell's 1984 was published. In it Winston Smith works in a government department that creates fake news (or propaganda if you wish). Today, fake news is rife and is considered, by some, to be a legitimate political tool. But at the time Orwell was derided for this view. Plus ça change, plus c'est la même chose
The thing is, can we prevent all this computerisation from happening? Probably not. As with climate change, we sleep walk into the future and the damage will be done before we even realise that we are doing damage. Then we have to try to play catch up, in the face of fierce opposition from vested interests, to try to rectify the situation.
On that cheery note, I’ll leave you to think about which work by a real live author you will next buy. Why not browse a while, while you’re here?