According to research from “The Atlas of Emerging Jobs” project, not only librarians, dispatchers, bookkeepers and archivists will have to find new jobs in ten years, but also representatives of professions that are regarded as creative: journalists, copywriters and translators.

The primary reason behind such changes in the global labor market is the technological revolution in general and, accordingly, the automation of many simple processes. As said by some enthusiasts, artificial intelligence can successfully perform the work of the fourth estate and some PR specialists. 

Today, AI is no longer just about “Alice” and “Siri” answering questions, or streaming services that recommend music, movies and TV shows to the user. Computer algorithms help HR specialists quickly process resumes when it comes to mass hiring for relatively simple positions, and marketers – to optimize expenses when conducting advertising campaigns.

Robot writers write news for major media, compile earthquake reports based on data from geological services, and even track the criminal environment by “working” with reporters.

In that sense, artificial intelligence is, on the one hand, a particular case of RPA (Robotic Process Automation), business process automation, and on the other hand, cost minimization. So should the word masters seriously fear that robots will replace them, and should media companies switch to using AI solutions instead of real people?

Robots: news is our profession

AI has already managed to establish himself quite well not only in writing simple weather reports but also in a variety of other topics.

The Wordsmith’s artificial intelligence system, using quantitative data analysis methods and the Amazon AWS cloud service, writes news for the Associated Press based on companies’ financial statements. According to the agency’s estimations, this saves up to 20% of journalists’ time.

Narrative Science Company has developed a similar Quill technology, which was tested in Forbes. Chris Hammond, the founder of Narrative Science, five years ago, stated that a robot writer, for example, by analyzing data on a variety of baseball games, is capable of imitating the unpretentious style of sports-writers and producing thousands of texts almost instantly.

Bloomberg uses its Cyborg system. Artificial intelligence has already prepared and published thousands of financial articles.

The Heliograf system used by the Washington Post has performed well in covering the US presidential election by creating newsletters.

Emails from Amazon

Phrasee technology based on Deep Learning and neural networks write emails to clients. The service is in use by Virgin Holidays, Domino’s Pizza, Groupon, eBay and other major companies. So if you receive letters from Jeff Bezos’ company in English, you may be reading a text written by a robot.

Companies are happy with the work of artificial intelligence: Molly Prosser from eBay believes that Phrasee is as crucial in copywriting as Adobe Photoshop is in working with images.

What can artificial intelligence not handle?

AI can free the author from routine duties, to be more accurate and faster, as is already happening in leading news agencies. A simple analogy can be made – autofocus systems save the photographer’s time and help them take a picture at the right time.

However, news can hardly be called ideal: robot writers manage their “duties” too formally. AI may give a headline to the text – “Company X increased profit in the first quarter by 25%”, but is unable yet to come up with the title “Company X surpassed competitors, increasing profit by a quarter.”

But machine learning algorithms are improving: it hardly takes very long for artificial intelligence to write better.

Intelligence is not consciousness

The main problem is that a machine can imitate human intelligence, but not consciousness: it has no emotions, memories, or past experiences. Therefore, creating an emotionally colored text full of details is beyond the power of robot writers.

Readers do not always need a simple statement of facts – except, of course, weather forecasts. A lively but straightforward single-lined post on social media can go viral, but a detailed note on the same topic usually does not.

Articoolo: personal experience

In our work, we sometimes face the need to write a relatively simple text in English quickly. The developers claim that the online service Articoolo can write and rewrite texts in six languages, distinguish between contexts, and understand that Apple is both the name of the company and the fruit. You just need to set the topic in a couple of words.

There were no illusions. From the very beginning, I was sure that a robot writer would not be able to cope with such a task independently. But I expected that I would get a text that would serve as a kind of “starting point” for the author or editor. Moreover, Articoolo automatically checks the result for uniqueness, which saves time.

Articoolo wrote a four-hundred-plus-character note about artificial intelligence in phenomenally fast time: in just 48 seconds. Excellent result. However, in the end, I got a story about the peculiarities of the intellectual development of children with autism. The second attempt also came to nothing: for some reason, the service was fixated on autism. I learned that “caring for such children requires patience and understanding.”

Yes, to the developers’ credit, the text was linguistically correct.

Are we going to use such services in our work? Certainly, not yet.

How to get the most?

Failed experiments of Articoolo are not a reason to consider the use of artificial intelligence for producing web content as science fiction.

It seems that robot writers will certainly replace copywriters who write SEO texts. However, this will not happen in a year, but it will probably happen in three to five years.

  • Many similar solutions will follow Articoolo. The situation should be monitored continuously; otherwise, more attentive competitors will start using such systems any day now.
  • With the arrival of new online services, the cost of texts is unlikely to be higher than current prices – $26 for thirty articles. A simple tip for copywriters and rewriters: pump your skills. Soon there will be no one on freelance exchanges to offer “unique texts” at bargain prices.