Back to Gartner for Marketers Blog

Content Marketing in the Age of the Robo Writer

By Jake Sorofman | March 11, 2015 | 2 Comments

If you saw The Imitation Game or if you’re at all familiar with the principles behind artificial intelligence, you probably know about the Turing Test. It’s a measure of a computer’s ability to convincingly imitate human intelligence—and it’s seen as a key hurdle in the intellectual contests of man versus machine.

Predictions of machines overtaking human intelligence are by no means new. But today they’re more than just the stuff of dystopian sci-fi fantasy or even longstanding debates on the role of IT in the age of cloud automation. Today, the machines are marching into all corners and quarters of business and life.

Let’s face it: automation happens. With every subsequent increase in scale and complexity we find offsetting answers in the form of machines. Marketing, it goes without saying, is hardly immune.

Consider, for example, the role that automation already plays in targeting, orchestration and measurement of marketing programs and campaigns. Or the role it plays in the trafficking of media and serving of ads. Or even in the collection of customer insights, where focus groups, mystery shoppers and field research may persist, but the voice of the customer has largely become the domain of machines.

Automation—a lot of it—is simply necessary to execute marketing at scale, particularly in a digital world.

But while we may willingly acknowledge the necessity of handing off largely operational marketing functions to machines, creative domains have always felt more or less invulnerable to algorithmic creep.

It turns out that’s not quite the case.

Today, machines are in relatively wide use to automate away the utterly ancient art of writing. Companies like Narrative Science and Automated Insights are helping to crank what turns out to be reasonable cogent copy, all without the procrastination, the hand-wringing, the pacing and the evasive dish-washing that afflicts many deadline-bearing writers of the sentient variety.

It makes you wonder whether yesterday’s scribblers could be replaced by tomorrow’s robo-writers.

It’s a shocking notion for actual living, breathing humans—like me—who treat writing as something of a sacred ritual. But it isn’t nearly black and white as some pro-robo doomsayers may have you believe.

The truth is that even the most advanced forms of automation are still reductive and rote. They may be able to turn data into words and phrases that, in combination, are coherent—even vaguely pleasant—to the human reader. They may be able to refactor these words and phrases to create new combinations.

But they’re unlikely to produce anything of original value. Save that part for the humans.

As automation creeps into the once invulnerable world of creative, it brings to mind a model for harmonizing the role of humans and machines:

  • Human effort = new ideas, unexpected connections, original reportage and creative renderings.
  • Machine effort = reporting of data and factoids, reductive words and narratives.

In combination, this could result is a substantial amplification in content output. And it turns out that amplified output is exactly what many marketers need to effectively feed the beast.

But before you tar me as pro-robo, keep in mind that I advocate for humans first—and I recognize that, in the imitation game, while machines may be closing the gap, there’s inimitable value in human output.

Fear not, content marketers. The machines may be here, but they’re bound to be at your service.

Comments are closed


  • Zeynep says:

    This is really interesting Jake – can you give some samples of robo-content? What are some scenarios where this type of syndication would be helpful? of course – automated translation such as Google, Duolingo etc is the first scenario in my mind.

  • Brad Zimmer says:


    Glad you weighed in here, and so soberly. I’ve seen the robo-content be effective in the conditions you mentioned – when the content is nearer a rote recitation of data points and the relationships between them (which in the end, I suppose, is just an equation first extracted and then reconstituted to text.) The “unexpected connection” and “creative renderings” you mention are the crux, aren’t they? I’d wager that until machines improbably become the primary consumers of content (or till humankind utterly lose its interest in the human condition), anything with even a hint of depth cannot be automated.