About a year ago, I realised that AI was about to get good enough to replace most copywriters.
Bad timing. I’d just turned 51. I needed to earn my keep for another decade and a half.
And there I was, pitched against machines that had read billions of books, could speak endless languages, could research so much faster than me, knew how to apply proven copywriting techniques to every post they wrote, and were even mopping up arcane things in which I actually had a weird expertise (14th-century chronicles, anyone?)
Yes, AI writing can be clunky. It keeps saying things like ‘sits between’, ‘the more defensible version’, ‘the real work is’, and ‘in failing, you’ve uncovered a more important story’. Oh, and ‘earns its keep’.
But when AI seems to be solving stuff like this every five minutes, it’s surely not beyond its ken to iron out a few verbal tics. Indeed, these days you can tell it what tics you want it to iron out.
But… don’t humans want to read human writing?
The better AI got at researching and copywriting, the more I noticed that many fellow commercial writers pitched the notion that human-written copy is better than AI-written copy.
I’ll give them this. It often is. But as copywriters, we’re putting the AI output under a pretty strong lens.
Most readers don’t. Indeed, most of them can’t tell the difference between human- and AI-written stories. They often rate AI stories as better than human ones. Unless they’re told a particular story has been written by a human. Then the results change.
As copywriters, we spot patterns that most readers don’t. And to be fair, a copywriting project costs a lot more than a Kindle Unlimited subscription.
Don’t build obsolescence into our profession
These days, if the main positioning statement we fall back on is being human, we will probably go out of business.
We write for price-sensitive sectors. If a smart person using AI produces well-researched, properly validated copy that (after some intelligent editing) converts, then many clients will buy that – rather than hire the copywriter who charges more for researching and writing much more slowly and more expensively.
It reminds me of what happened to print firms in the early 2000s. Cheap digital print almost killed UK litho. Translators suffered a bit later – who would employ a translator to do a job from scratch when they could start with a surprisingly good DeepL.com translation and tweak it?
(By the way, I have put years of old-school work into my French. And DeepL helped me make it better. Is it not at least possible that we can learn, as writers, from AI?)
Someone who truly understands AI can offer robust, validated research. Quickly. They can use it as a sparring partner for new ideas. Sometimes less quickly. They can use it to improve rather than degrade their writing abilities. In short, they can use their existing skills to direct it as a conductor learns to direct an orchestra. If they really work at it.
Our offer has to change
When I finally realised that AI was set to replace many of the usual copywriter skills – the ability to assimilate information, the instinct to uncover a buyer’s needs, the gift to draw on every published thought of people from Claude Hopkins to Ogilvy, Dave Trott to Seth Godin, from… then I realised I couldn’t compete.
Unless you want to become one of the few niche ‘human’ writers offering consumer-facing work in 2028, you probably need to change your offer.
That doesn’t mean throwing away your industry expertise. It means getting to understand how AI works and how it develops – and then making the most of it.
I’ve been using it to give me enough aggravating feedback and lateral thinking to write a novel. The arguments have got my words down on paper at last. They’re mine.
I’ve been using it to vibe code – my own website, my content audit software, and even a very beautiful (upcoming) WordPress site for a Catholic church I am part of.
And I’ve been using various AIs to take a long, hard look at themselves, their work (and that of their AI chums). It’s resulted in a white paper I’ve spiked, an insight into how some AI clients compile their recommendations, and at each turn something new I can tell clients from experience and authority.
Such as why it’s fairly easy to get recommended by Perplexity, but why you have to do different stuff to get noticed more by Claude and ChatGPT (I’m keeping this research to myself for now).
Personally, if I’m in a room with a new client – one who’s more interested in learning how to get business from AI than preventing their copywriters from using it – I’d rather have some original research at my fingertips. These days we need to sell ideas, not just copy.
That’s why it’s the extra stuff we can offer with AI, followed by the writing (however it’s honed), that’s going to save us.
Disclaimer: this article was written by me. I know my audience… bung it through Pangram or whatever…