In June 2017, the Oxford University Future of Humanity Institute and a team from Yale University published a report which could be considered grim reading for (copy)writers. Entitled “When Will AI Exceed Human Performance?”, the paper surveyed 352 experts in the AI field, asking them whether they thought AI would surpass humans for a range of activities. The consensus was that within ten years, AI bots would easily be able to create passable high school essays and that in just over 25 years, the New York Times bestseller charts could be full of AI writing. So, you might think that humans should lay down their pens. Surely, within a few years, computers will be writing blogs, telling stories and capturing the attention of billions. But we don’t think so. At least not quite yet. While auto-generated content is here to stay, humans are going to be instrumental in shaping what we read for the near future.
Auto-generated content is anything that has been produced solely by software and is subsequently published on the web. This isn’t a new phenomenon. In the past, controversy has raged about “article spinners”, which take existing texts and automatically rewrite them, often with poor results. But there’s a difference between the promise of AI auto-generated content and primitive spinning tools. The current generation of AI writing tools is much more sophisticated – and these tools are improving fast. In Japan, readers got a taste of this in 2016 when a computer-generated novel passed the first screening phase for a major literary prize.During the 2016 Rio Olympics, the Washington Post relied on a tool called Heliograph to compile simple reports about medal-winning performances. In both of these cases, auto-generation software used algorithms with certain rules to create intelligible text, creating very different styles of writing in the process. As you read this, major news organisations like Reuters are investing in AI to make the use of these algorithms (to write daily reports) mainstream.
However, the question is: are AI writers going to reach the level necessary to put human (copy)writers out of work? A consideration of AI’s current capabilities suggests that mass unemployment is a fair distance away. When you look ‘under the hood’ at a basic content writing bot, the digital surface sometimes exposes an analogue core; these AI tools are not infallible. In California, articles written by Artificial Intelligence took USGS officials back to the 1920s when they alerted locals to an earthquake which took place in 1925. The Associated Press uses a tool called Wordsmith to write up financial announcements such as corporate earnings, which seems like a pretty simple set of information to interpret. But even then, humans need to create the right data sets and guide the content writing bot. Even this isn’t something AI can do completely unsupervised.
Even if the articles produced by AI bots are works in progress, the trajectory of progress is clear. For relatively simple, routine tasks like reporting on a baseball match score or the profits of a company, bots will soon be delivering what you read. But salvation could come from an unlikely source. Thanks to the way search engines are changing, successful content has to work harder to connect with readers. Ironically, the better Google’s algorithms become, the more creative and professional content writers have to be – and the harder it is for AI writers to compete. Marketers need human writers to inject empathy, humour, and style into the texts they publish. Potentially, this is leading to a two-tier content scene where AI does routine work and humans produce elite material. In theory, companies could use AI to free up resources for the creation of more in-depth, human-made texts. If that future materialises, human copywriting solutions will be more vital than ever, not less so. And it’s a future we’re comfortable with at greatcontent. So, if you want to be part of it, get in touch and order some great (human) content from us.
Text: Sam Urquhart