The rise (and fails) of report robo-writers

In the 1991 blockbuster Terminator 2: Judgment Day, the time-travelling Terminator (Arnold Schwarzenegger) reveals in deadpan tones how artificial intelligence (AI) put in charge of America’s weapons will deliberately trigger nuclear Armageddon.

‘The system goes online August 4th, 1997. Human decisions are removed from strategic defence. Skynet begins to learn at a geometric rate. It becomes self-aware at 2.14am Eastern Time, August 29th. In a panic, they try to pull the plug.’

The film is a turbo-charged slice of Hollywood hyperbole. And even 26 years later, relatively few people believe computers are in imminent danger of threatening our very existence.

But it’s a fact that AI is transforming businesses worldwide – and if your job description involves writing reports of any kind, it’s time to sit up and take notice.

What are AI report-writers? And could they make you redundant? @EmphasisWriting investigates. Share on X

The warnings certainly sound dire enough. Report writers feature in a list of the ‘five jobs robots will take first’. And, in fact, so-called ‘robo-writers’ already produce billions of pieces of content each year. Leading research and advisory firm Gartner predicts that by next year machines will author a fifth of all business content. So what are robo-writers? Are they any good? And are they about to make you redundant?

What’s the threat?

Let’s first consider what human report writers are actually up against.

‘Robo-writers’ is something of a misnomer. Of course there’s no army of droids – like a sea of C-3PO lookalikes – sitting awkwardly in front of a bank of computer screens, clunkily tapping away with metallic fingers on their keyboards.

Put simply, ‘robo-writer’ content is the product of software that uses natural language generation (NLG), a branch of AI that generates language as output on the basis of data as input. And significant recent advances mean that it can now write surprisingly readable reports, often with an authentic-sounding narrative.

The heavy emphasis on data in the world of finance has made it a prime target for NLG platforms, but robo-writers are producing content in every business sector for everyone from sole traders to multinational corporations. Big-name clients include the likes of Yahoo! and Samsung, and last month Coca-Cola announced it wants to use AI bots to create its ads.

And the evolution doesn’t end there: they have morphed into robo-journalists too. Associated Press (AP), for example, has partnered with US-based Automated Insights to produce quarterly earnings reports for about 4,000 companies. Previously, its journalists could only manage to publish these reports for around 300 firms.

Writing, but not as we know it

Compared with their human counterparts, robo-writers are incredibly productive, creating thousands of articles or reports per second if required. In 2015 alone, Automated Insights used the algorithms in its Wordsmith software to create more than 1.5 billion pieces of content for its customers.

That points to another huge plus of robo-writers. Organisations can use AI to produce highly personalised, detailed updates or reports for large numbers of people. For example, a financial services company in the UK might respond almost instantly to the words of our chancellor by sending its clients personalised emails titled ‘What today’s Budget statement means for you.’

So should we forget trying to improve our own report-writing skills, and simply prepare to let the bots get on with it?

Well, no. In fact, absolutely not. And that’s not because robo-writers get lots of things factually wrong. It happens, but that’s usually down to errors by the humans writing the algorithms. Rather, it’s because people still beat the machines hands down in some key areas – at least for now.

Analyse, report, repeat

It’s not a cardinal sin, but robo-writers do appear prone to needless repetition – irritating when you regularly read content provided by one particular platform. (I searched for the phrase ‘generated by Automated Insights’ together with ‘The results exceeded Wall Street expectations’ under News in Google, and I got 469 hits.)

AI-generated content can also lack perspective, all-important when analysing data. It may be true, for example, that drinking red wine reduces your chances of catching a certain virus by 50 per cent. But if it turns out your risk of falling foul of the virus was 0.002 per cent to begin with, and now it’s 0.001 per cent, then drinking red wine really makes no tangible difference.

AI bots create thousands of reports a second. How can human writers compete? – @EmphasisWriting Share on X

So we’re often better than bots at setting things in context, and we understand the need to get the nuances right, too. (That’s one reason why long lists of bullet points in a report can be unhelpful – the reader can miss the subtle, but important, links between apparently unconnected pieces of information.)

What about trust? So much of what we do in business is about building relationships, based on empathy. Even when writing a formal document such as a bid or proposal, we boost our chances of success when we show we truly understand the prospective client’s issues. And as a client, isn’t it more reassuring that the author of a document is someone whose name you recognise and who you believe in, rather than the name of a software platform?

Partnering with a bot

For the most part, human report writers remain indispensable, but it is staggering how far this technology has come. It’s proving hugely powerful for companies of all sizes around the world, especially for shorter, data-rich reports. Companies should grab this opportunity to offload the least-rewarding writing tasks onto a machine, freeing up report writers for higher-level writing.

Even then, there’s plenty of scope for them to work in partnership with robo-writers. Take AP – it ensures earnings stories about top-tier companies such as Apple and Google are written by journalists, who can incorporate AI-generated copy if appropriate.

Unlike the dystopian world of the Terminator movies, it’s not a case of ‘us’ or ‘them’. With the arrival of Big Data there’s simply too much information for us to communicate without the help of robo-writers. We’re only human, after all.

Generated by Doug Nel

Image credit: Ociacia / Shutterstock