AI and Copy: It’s Beyond Words

AI and Copy: It's Beyond Words

I’m a writer.

And a nanoscientist. And a punk. And a pet-loving six-year-old. 

I’m SoCal one minute, Silicon Valley the next. Yesterday was 1992. Tomorrow? It’s 2032.

What I mean is, copywriters wear an awful lot of hats. We have to, or else we’d be using the same voice for trainers as we use for toilet cleaner. 

But when it comes to having hat game, one thing seems to have us beat. Sure, I’ve got a row of lovely lids to wear, but AI has the whole damn millinery. At least, that’s the narrative everyone’s running with.

Who wore it best: me or Claude?

In the design world, conversation around AI has gone in every direction imaginable. Experimentation and adoption are commonplace. But among copywriters? It feels like there are fewer perspectives, which is odd considering we’re paid to have a lot of voices in our heads.

On one side, we’re championing the lines that AI can’t write, trying to reclaim our beloved em dash, and doubling down on our own point of view. On the other side, we have the mavericks: the small percentage who claim to have trained AI to write exactly like them, with a scary degree of accuracy.

It’s the kind of stuff that has copywriters clutching their pad and pen, asking “there’s got to be more to all of this, right?” And there is, once you take AI out of the Google Doc. 

The thing is, most copywriters would agree that AI isn’t going to change the words we write. But it could change how we tackle some of the challenges that come with being a writer. Like designing your own header image for an article about copy and AI.

And that’s where it gets interesting.

We set the tone, can AI help them stick to it?

Take tone of voice for example. You spend weeks chipping away at a verbal identity that fits that brand like a glove. Then comes the inevitable feedback: “This is great, now how are we going to make sure that all the non-writers can use it?” You suggest a workshop, and it works well—to an extent. Eventually, copy starts to come out in the new voice. It’s not quite how you would do it, but you can’t be in two places at once. Tale as old as time.

It makes me think of Jasper and its Brand Voice tool. To put it simply, you build its knowledge base (or ‘voice’) with pre-written content and principles, then ask it to write and review new content using that voice. “AI that sounds like you,” says the shop window, but in practice the results are mixed—and that’s after the massive investment in time and money to set up. 

For now.

In time, AI could untie this knot for real, allowing non-writers to comfortably and consistently write in one brand voice. Copywriters will still be crafting that voice (along with the more creative brand messages) but it’ll be embedded with more ease. Now that’d be something.

Once more, with feeling

Then there’s the challenge of getting the written word sounding right when read aloud. Let’s say you’re working on a rough script. You want to turn around a quick voice-over to make sure it’s word perfect before investing in a professional voice-over artist. Those text-to-speech models only get you 50% of the way, giving you the right tone but the wrong rhythm. You press play, and you’re cringing at your own words. Been there.

Hume’s brand new voice conversion tool means we no longer have to rely on our friends to do test reads. Just record your own script at your own pace, then apply the right AI voice. Down the line you’d want a human voiceover artist to perfect it, but for proof of concept it’s a game-changer.

Staying scrappy, staying human

Both of these scenarios point to a future where copywriters still have their way with words. A future where AI’s not helping us put words on a page, per se. Rather, it’s helping us get our words out into the world in the way we always intended.

As I write this piece I’m listening to LUX, the audacious new album from Rosalía. Over its 49-minute runtime, she sings in 13 different languages across opera, flamenco, hip-hop and pop. That’s a hell of a lot of hats.

How did she do it? As she tells it, she first ran her original Spanish-language lyrics through a translator on her phone, then took those sketches to native speakers to sharpen the meaning and pronunciation until it was perfect. In other words, scrappy digital ideation, considered human execution. 

As for the London Symphonic Orchestra, heard throughout the record? Well, their parts started as digitized string arrangements, which she took to a composer before finally investing in recording sessions with the LSO. Orchestras are expensive, after all.

LUX has been praised for being uncompromisingly human, and will likely end up being the year’s most critically acclaimed album. But there’s something in the fact that on the long road to get there, technology played a small but crucial part. 

It’s a dance we’ll all be doing, in our own way, if we want to hold on to our hats.