When AI starts writing thought leadership

By Rachel Ainsworth.

I have to confess to having described, in some of my more confrontational moments, pieces of thought leadership as something that could have been produced by “any business-savvy intelligent person, locked in a cupboard with a desk and an internet connection”. However, having read Man v machine: can computers cook, write and paint better than us? in The Guardian’s weekend supplement, it seems the “person” element of my damning description is now defunct.

Having pitted IBM’s Chef Watson against ingenious chef Yotam Ottolenghi, the Painting Fool against established artist Jane Moon, and Google Translate against experienced humans, The Guardian provided artificial writer Wordsmith and its own football writer with a table of facts about recent events in the English Premier League.

In all cases the machines performed admirably, but not quite as well—at least in the eyes of a human audience—as the expert humans. For me, the paragraphs created by Wordsmith brought to mind many pieces of thought leadership I’ve found myself reading recently. No, not because consulting firms have started writing about football. But because, like many of the publications I review, Wordsmith’s output makes sense, offers facts and figures, and even contains the occasional quote. But it lacks the flashes of insight and engaging tactics that set the best thought leadership apart from the rest.

So what can we humans do to raise our game above that of the artificial writer?

1. Machines—and possibly even business-savvy intelligent people locked in cupboards—can find data to support or refute hypotheses, but they need guiding on their way by experts with hunches about what to look for. What are your hunches about what really makes a difference in your area of expertise?
2. People relate to people. Does your content come from a faceless organisation or a likeable human? Do readers know who you are, where your passions lie, and why they’d like to spend more time with you and your words?
3. Humans—for now at least—can outperform machines in terms of telling engaging and memorable stories, or using surprising analogies to help their audience understand a topic. Are you doing enough to engage with your readers? Or are you simply presenting facts, figures and occasional quotes?

If you’re not planning to do any of these then you might want to consider saving yourself a whole lot of trouble by finding a machine that can write your thought leadership. But don’t be surprised if few humans make it past the first page.