‘Automated Writing’: Implications for Digital Communicators

‘Automated Writing’: Implications for Digital Communicators
12. November 2019 Falk Rehkopf

‘Automated Writing’: Implications for Digital Communicators

'Automated Content Production' and its implications for digital communication experts

AI-driven technology continues to evolve at a fast pace and, according to data from the World Economic Forum, the global AI industry will expand by a staggering 50% every year over the next five years and is forecasted to be worth $127 billion by 2025. However, when we talk about AI-driven automation, most of us tend to think about industrial robots used for manufacturing, self-driving cars and even vacuum cleaners. But, there is one process that has already been automated much more than many of us realise: content production.

In a nutshell, automated content production – or automated writing – refers to when AI-based technology analyses and interprets a data set and runs it through an algorithm to create a human-readable text (it can also refer to ‘AI auto-completed texts’). This automated production of content can be useful in a variety of contexts: essentially, any industry that has the need to take a data set and generate text out of it can put this technology to good use. It doesn’t come as a surprise, therefore, that Narrative Science‘s chief scientist Kristian Hammond expects that within just 15 years 90% of all articles will be written by AI.

But, what will be the implications of the looming widespread adoption of these new technologies for digital communicators?

Automated Journalism

The journalism industry may have the heaviest need for AI-driven, automated writing. As journalist and communications professor Charlie Beckett puts it: “AI, in its broadest sense, provides all sorts of opportunities for journalism“. One of the benefits for news publishers is the fact that automated writing allows them to expand their reach and write about topics that they no longer have resources for – such as local news stories and sports.

In 2016, natural language processing (NLP) and technology company Automated Insights generated over 1.5 billion pieces of content, allowing publishers to also report on financial news and other data-driven journalistic content – at scale. In the same year, Heliograf, the AI writer of the Washington Post, created more than 500 articles about that year’s election results. For comparison: in 2012 it took four employees 25 hours to put together and post just a small part of the election results manually.

Another example of automated journalism comes from the newsroom of Swedish newspaper Aftonbladet. Through a collaboration with Swedish tech company United Robots – which develops bots to automate content production – the newspaper increased not only their scope of coverage but also more than doubled their journalistic output.

But efficiency aside, AI-based content generators can also be of support to writers to find creative inspiration. This is what, for example, Forbes’ Bertie does: It learns from different writers and tailors its output for them. It also recommends topics to contributors based on their earlier content pieces and suggests headlines and images that resonate with the respective article’s sentiment. Since deploying Bertie, the site has increased its traffic and doubled its number of monthly visitors.

On top of that, AI-generated news production and presentation reached another level at the end of 2018, when Chinese Xinhua news station has created ‘humanoid reporters’. These ‘AI presenters’ resemble humans who sound and look like real-life newscasters:

Automated Marketing

The same technologies used to generate journalistic content also have many applications in marketing. In particular, content marketing has become increasingly relevant for many brands but has now reached a tipping point as the amount of content produced is far greater than our ability to consume it. As a consequence, optimising content to achieve sufficiently high engagement rates is now paramount for communicators.

One way to do this is by implementing personalised content-based strategies. And while it would be impossible for brands to hire humans to write personalised content for individual customers, AI technology can actually deliver that. As an example, the UK newspapers The Times and Sunday Times were able to reduce newsletter unsubscribes by 50% using an AI-curated newsletter which personalises content based on subscribers’ individual interests. The production and delivery of personalised content is enabled by content automation which, in turn, is fuelled by ‘Content Intelligence’.

Automated PR

Just like in marketing and journalism, automated writing can also support PR pros to engage with their publics more effectively. As Rebecca Wilson – Executive Vice President, Singapore & Australia at WE Worldwide – puts it: “Those who embrace AI will have a leg-up in connecting brands with new audiences and on a deeper level, providing more value to organisations”. 

While AI can help PRs, for example, by automating menial tasks such as scheduling emails, social posts and more, it can also help produce data-driven content. With automated writing, PR teams can deliver crucial content to update people swiftly and efficiently – of particular importance in times of (PR) crises. Already today, PR agencies offer automated press releases and automated performance reports as well as other AI-driven services.

Ideas and technologies around the concept of ‘Conversational PR’ take automation to a whole new level. Conversational PR makes effective use of technologies such as chatbots and voice assistants to engage publics in automated, yet personalised conversations about a brand, topic or an organisation. This approach is used for information and crisis management as well as automated media relations.

In addition, tools such as SumoStory, for example, elevate efficiency levels within the PR pitching process: Algorithms match journalists’ interests with those of organisations to only send relevant content to journalists – automatically.

Benefits of Automated Content Production…

While many communicators fear that AI-driven, automated content generation will render humans obsolete, we agree with Radar AI‘s editor-in-chief Gary Rogers who says that “while this has been true in most industries and may happen in media, there is a broader picture of AI’s enabling rather than employment-destroying qualities. AI can take over repetitive and boring tasks, which frees journalists to do more important work”. Interestingly, US news agency Associated Press estimates that AI can improve news accuracy and helps to free up 20% of reporters’ time – which they could use for more interesting and impactful work.

Another benefit is that it could bring about higher levels of objectivity: while human opinions are prone to bias and limitations, machines are not – provided they are fed the right data. This could also decrease the occurrence of unethical framing and counter-framing practices, re-establishing higher levels of trust as a consequence.

On top of that, AI cannot only produce content but also automatically detect phenomena such as fake news. As an example, the neural network Grover can detect such news with a 92% accuracy rate.

… and the Risks

Today, machines lack the soft skills required to grasp the nuances of human communication and to discern truth from fiction. An improperly trained AI-automated writing software could easily turn into a copy of Microsoft’s AI-powered Twitter bot Tay, released in 2016. Tay was designed to mimic Twitter users and have conversations in real-time but within just one day, Tay was taken offline because it was tweeting racist and xenophobic rants. 


To address and fight this challenge, the ‘Partnership on AI‘ initiative has, for example, formed a new committee, made up from organisations with a focus on AI, other technologies as well as media companies. The declared objective is to develop software that exposes false information and at the very same time authenticate trusted news media sources and content.

Ethics Also Matter in Automated Content Production

Experts rightfully worry that AI content production can be exploited by malicious individuals, for example, to spread false news. Others believe that algorithms are already biased as certain opinions are already embedded in the underlying math.

Also, who will, for example, take the blame in a lawsuit based on something an AI wrote? The way we see it, humans still have to be responsible and liable for AI-generated content – which would also require that experts constantly monitor these machines. Rather than being substituted by machines, humans would acquire different roles and responsibilities within content production processes. Therefore, communicators should:

  • Provide data accuracy: When the story is written completely out of data sets, communicators need to question the validity of the data itself beforehand. Is the data accurate and reliable?
  • Apply human judgement: AI technology cannot yet decide what is appropriate to write about. Today, humans still need to take on the responsibility to check and approve the produced content.

AI at Ubermetrics

Ubermetrics‘ Content Intelligence platform is, already today, driven by AI technology. However, in 2020 our platform will include features that enable communicators to automatically reply to questions people might have about a specific piece of content. Ubermetrics also works on an auto-complete function which will finish texts based on a topical input. This will increase productivity and may also improve the customer experience.

In addition, Ubermetrics is proactively engaged in several research projects (such as QURATOR and others) that look to develop AI-based features to help communicators to curate and analyse digital content more effectively, including AI-based text summarisations.

Ubermetrics Content Intelligence Powerhouse

Outlook: Embrace Change

While many consider automated content production – or AI in general – a threat, we recommend to see it as an opportunity for advancement. We agree with communication expert Christopher Penn who says that “the use of AI is not meant to replace PR professionals’ jobsinstead, it allows us to do our work at scale”.

The same adage goes for marketing and journalism. AI writing, in its current form, can’t replace journalists or writers because there is more to writing than just turning data into content. Machines still lack the ability to understand nuance and context.

While some businesses and news organisations are already taking advantage of this opportunity – or are at least trying to understand it – the majority lacks behind. The way we see it, communicators need to fully understand what AI-driven automated content production entails to avoid falling behind – but, more importantly, realise the full potential in putting these innovations to productive use.

Chief Marketing Officer