Will the Next Great Novel be Written by a Computer?

Will the Next Great Novel be Written by a Computer?
The Exela Blog

Will the Next Great Novel be Written by a Computer?

Share This
by Matt Tarpey

The words ‘artificial intelligence’ often bring to mind lofty visions of a miraculous, futuristic world. But, despite its name, AI isn’t really all that intelligent. Most current AI works less like a human brain and more like a complex system of pattern recognition that relies on set inputs and desired outcomes to operate.

Still, today’s AI can do impressive things based on pattern recognition. Some might even say AI is already capable of being creative. For instance, AI is able to compose music quite impressively after being trained on a set of MIDI files gathered from various genres. It views the data over time to “understand” the context that the song’s melodies fall into and predict the next notes in a sequence. After millions of rounds of a game of hide and seek, AI systems managed to discover creative ways to win – including exploiting programming errors and glitches. AI is also making significant strides in the field of writing.

As exciting and interesting as these developments may be, many people have raised concerns about the potential for smarter automation technologies to replace a great deal of physical human labor, and so destroy manufacturing and service jobs as it goes. Recent developments in AI might cause knowledge workers and creatives to take note as well.


The Expansion of AI Into Writing

The current status of AI-based writing can be seen predominantly in areas where the “creative” aspect is minimal - or where the writing is more of a reporting of the facts than an artful example of storytelling. You might be surprised to hear that The Associated Press and The Washington Post, for instance, already use AI to write news articles and many organizations are using similar technology for financial reports. Subjects like these are easier for AI because the system just needs the right facts and a style guide to create an easy-to-read piece.

Still, there are rules and patterns to good storytelling - just take any creative writing class and you’ll begin to recognize the formulas in all sorts of entertainment, from classic novels and movies to comic books and TV shows. The basic structure of a good story is something that can be codified and programmed into an algorithm, and the elements of language and stories are already pretty well-defined, as are their relationships in good writing. As AI advances, input from creative human minds may become less critical to the production of “creative” work.

Researchers are also completing more ambitious projects than simple pattern- and statistic-heavy articles and reports, though often with less convincing results so far. While computers have produced poetry that could pass as human-written, as well as a whole Jack Kerouac-esque novel that came surprisingly close to the real thing, the output is generally created one sentence at a time, leading to an awkward, jerky flow, and inexplicable, confusing, and even nonsensical narratives. 

AI may not be able to create a more satisfying ending to Game of Thrones (low as that bar may be) quite yet, but its understanding of the characters within these stories is a strong indication that these attempts at creativity have the potential to become more advanced.


What Will AI Mean for Professional Writers?

The future of creative writing isn’t wholly dependent on whether AI can churn out competent writing that gets the facts right — it also depends on how good the writing actually becomes, and perhaps more importantly, how expensive intelligent systems are to produce. If a publisher can use a bot to spit out usable content at a lower cost than paying a staff writer, then the majority of writers are all but unemployed.

For humans, however, the good news is there is currently no indication AI will ever be able to truly cross the barrier between copycat creativity and true imagination. AI isn't really "aware" on a deeper level the way a human is, and therefore doesn’t make some of the connections a human might. Even the most advanced AI still just consists of pattern recognition and big data lookups. If given a random piece of writing, it wouldn’t be able to tell you whether it was “good” or not - just whether it bore similarities to existing works its been told are “good” writing. True creativity requires much more than just recognizing and extrapolating patterns.

Making an AI system that’s a truly creative writer would require getting much closer to solving "the hard problem" of consciousness. This is what cognitive scientists call the problem of coming up with a tenable view of what conditions are required for there to be an “experience of being.” Genuine creativity likely requires this level of consciousness, and there’s currently no clear path to it in AI.

If AI stays stuck in an unconscious state, then, at the very least, the top tier of creative writing is still safe from the machines, even if much of the relatively mindless, fact-based work dries up. The problem is, even in this kind of scenario, the future doesn't look too rosy for the majority of writers who aren't the next Leo Tolstoy or Virginia Woolf.


What a Future of AI and Content Writing Could Look Like

There’s no point in sugar-coating it: a future with creative AI will be a hard one for content writers. It's already hard for most professional writers in an increasingly saturated market. The growth of freelance work is outpacing total workforce growth by three times

Add machines to the mix, and the writing market only becomes more saturated. News and media will be flooded with content from humans and machines alike. Books created by AI will stop making news because they’ll become so common. The line between which content comes from humans and which from AI will blur even more, making it difficult for anyone to stand out.

 Still, a world where AI programs dominate the field of content writing is a ways off, if it comes at all. The development of revolutionary new technologies is often met with concerns about its impact on jobs. By definition, automation is designed to take certain tasks from humans and hand it over to machines. It changes the nature of work, and often leads to new jobs that compensate for its own weaknesses.

In this way, AI is no different than automation technologies that came before it. A day may come when AI is capable of writing a compelling narrative touching on complex themes, with thought-provoking allegories and surprising twists that enrich the story, but it’s nowhere near there yet. For now, at least, AI can help improve a writer’s content, but it can’t create great content on its own. 


Share This