How AI is changing writing and careers in content.

There’s been much speculation about how artificial intelligence will change the way we work. No doubt, any work that can be easily automated is an opportunity — or target, depending on your perspective — for AI to replace human-powered jobs.

The work of writing, producing, and curating content is no exception. Algorithms are quite competent at writing coherent, readable prose and curating content. Taking my preferences, likes, searches, and other signals to curate personalized content in a novel way is what AI does best — when it works.

Robots can write good

Just about everyone using the internet has read or consumed content created or curated by robots. A few easy examples:

Yahoo! Fantasy Football recaps? That’s a robot.

Your weather app forecasts? A robot writes most of them.

That Spotify Discover Weekly playlist that significantly improves my commute? Love. That. Robot.

AI is very good at taking data-based content and transforming it into a narrative. That’s why the Associated Press started using it in 2015 for automating quarterly earnings reports — and “increased output by tenfold” — while also reducing the drudgery of formulaic writing for professional writers.

Will a robot write the next great American novel?

But just because something can be automated doesn’t mean it should be. It’s early days in AI, so the desire to show what can be done far exceeds the common sense to ask whether something should be done. For example, robots can write novels, potentially even good ones, but what’s the real upside to automating creative writing, just to prove you can? Is novel writing so expensive an endeavor that companies would be willing to develop technology to automate it?

Maybe. Once you’ve trained a bot, it will repeat the work for less cost than, say, a human author. But training a bot is hard, painstaking work. It takes time, technical expertise, writers, and a substantial financial investment to do it right and well.

In my experience working on customer care chatbots, you not only need technical talent to build the bot, but you require skilled, experienced writers who know how to create personality, voice, and a conversational experience that’s helpful, smart, and efficient. The latter is often an afterthought — and an additional cost that’s not factored into planning.

Robots aren’t that funny

Right now, it looks like the technology that powers NLG (Natural Language Generation) leaves more creative kinds of writing for human creators while it produces the less interesting work of writing recaps, white papers, and monthly earnings reports. As journalist Matthew Lee notes of Wordsmith, the product used to generated Yahoo! Football recaps:

“Wordsmith, it seems, is indeed capable of being snarky. But its jokes have been dropped in a spreadsheet by data scientists and writers. It still needs humans employed to write zingers. What would save publishers even more money is software that can write its own jokes.”

And that’s the part of automation that most customers and audiences don’t realize: content that has personality, snark, and insight is probably authored by a human. Most bots and personal assistants like Siri and Cortana are a composite of human-generated language — databases full of potential responses to commonly asked questions or prompts — and a layer of refinement, perspective, and personality provided by an editor. Ask Siri to tell you a joke and she’ll tell you one a human wrote. Ask Cortana to tell you “something spooky,” and she’ll give you a writer’s well-crafted answer.

Writers wanted

So while AI is transforming technology, it’s also changing content. And, as forms of content experience evolve to include modes like conversation, voice, and other ambient technologies, so too will the skills needed to write for this future. The way I see it, there are a few emerging paths for work in content:

Bots

As bots become personalized brand agents and smart assistants capable of building relationships with customers, scriptwriters, brand content managers, and strategic communicators will be needed to deliver brand personalities that look, feel, and sound like the companies they represent.

Bots will also get smarter and conversations more authentic, thanks to human trainers who will teach and model conversation and empathy. Who better to model that than writers, particularly writers who have experience with dialogue and scriptwriting?

Voice

Most content written for the web isn’t ready to be voiced. How will experiences designed for static screen reading transform for voice? (Hint: text-to-speech technology isn’t a solution unless you want to sound like a robot.) Add to this that there aren’t enough scriptwriters to rewrite the web. New skill mixes are needed — voice actors, sound designers, content writers — to tune and train the translation from online to voice.

Place

What happens when traditional UI actions — ’start,’ ‘close,’ ‘buy’ — transform in physical locations where sound, signs, and haptics reshape interaction? Cashierless retail environments like Amazon Go use content in the form of QR codes, security and privacy information, badges, branding statements, and more. New skill composites are needed for these environments — physical store retailers, ambient designers, cartographers, privacy and security experts — to communicate new interaction models.

Explainers

Who’s going to explain how AI works — or doesn’t work — when systems learn and decide on their own? There’s a gap between the technology, the technologist who can explain what’s happening ‘behind the curtain,’ and the lay audience who needs this information. And while there are limited tools that can reveal how AI systems work, highly skilled humans will still need to provide levels of interpretation for different audiences. You can find that rare skill set among technical writers who can balance the business need for clarity with an expert understanding of how AI systems work.

Write what you know

In short, employability and career opportunities will go to those who can turn their skills to align with trends — and express the value of those skills in terms of how they help technical services connect deeply with customers. That means rethinking positions like “I’m a writer, not a machine” when you’re asked to train a customer support chatbot. As a “words person,” you understand more deeply than machine learning experts how humans talk, interact, and communicate. More importantly, you know what a customer wants from that interaction — understanding, empathy, and great service.

I’m a managing editor. I lead a team of content creators who are in the business of language — creating it, editing it, analyzing how it performs. Like most content teams, we have a style guide that sets some guardrails on how we express our brand, including things like punctuation and the words we use. It’s an organic document, evolving with the brand and with language.

Election season provides an enjoyable bit of theater for most voters, but it’s especially fun for people who work on language. Every four years, we can count on a fresh lexicon of phrases, expressions, and unfortunate responses that characterize the candidates and the tenor of the race.

Remember 2012, when Romney’s “binders full of women” entered the conversation? Or when Biden told a Virginia audience, at least half of whom were African-American, that Romney’s policies would “put y’all back in chains.”

Exactly.

I think we can all agree that the 2016 election has pushed the envelope on propriety — if not the number of times the word “pussy” ran in anything other than the back-pages of most city magazines — or a porn search.

I for one never thought I’d have a work conversation with another editor about whether we should reference Melania Trump’s ironic choice to wear a pussy-bow blouse to last Sunday’s debate in our weekly trends quiz. I mean, how can we not talk about what so many flummoxed newscasters tried so hard not to say?

There was a conversation about the rightness of this. Were we going low when we should go high? Were we just perpetuating more “locker-room talk” instead of letting the conversation move on to presidential candidates doing presidential things — like inspiring and persuading us of their fitness to hold the highest office in the land?

In the end, we took the middle road, opting for Ken Bones over pussy-bow. It was the right choice for the brand — it sidestepped the crude in favor of the careful. It also saved us an exhaustive debate over whether to use the hyphen: Is it pussy-bow or pussybow?

For the moment, neither is in our style guide, but with several weeks left until the election, I’m not ready to call “pussy” out.