“Haven’t you been replaced by ChatGPT yet?”
As a copywriter, it’s undoubtedly the question I’ve been asked most in the last few months (apart from “how’s the puppy doing?”) And when I assure people I’m still gainfully employed, the follow-up is invariably “aren’t you worried about it?”
In all honesty, no – at least, not yet.
There is no doubt that this new generation of natural language AI tools is a huge leap forward. Output is typically good and (mostly) grammatically accurate. Plus, it’s lightning fast and cheap – terms that haven’t been applied to most copywriters for several years.
Man vs machine
There’s equally no doubt that the next generation will be smarter, more fluent and just as fast and cheap. So how can we humans compete?
At this stage, I’m not sure we need to.
There are plenty of jobs where ChatGPT, or some of its less-heralded rivals, may well be the perfect answer. If you need to generate a quick product description, a web page, maybe even a few headline ideas, then it’s probably up to the task.
But there are plenty more jobs where – from what I’ve seen – natural language AI might struggle.
Garbage in, garbage out
Like any computer-based solution, effectiveness depends on the instructions you give it. In marketing terms, that’s the brief. One of the skills a human copywriter brings is to interpret the brief: ask questions, spot any gaps or inconsistencies, and make suggestions. Once a client knows a writer, and the writer knows the client, that briefing process can often be accelerated to little more than a two-line email. There’s a shared understanding of what the output should be and the writing style required.
ChatGPT can’t do that, so it requires a more detailed briefing – or series of directive prompts – every time. Are busy marketing managers always ready to do that? Yes, you can iterate or refine your instructions to focus the initial results, but this might begin to grate, as you tweak them for the fifth time.
All about style
Then, there are questions of style. Leaving aside the entertaining, though ultimately useless, requests to write like a pirate or your favourite Sesame Street character, the output I’ve seen from AI tools typically conforms to solid marketing principles and a ‘salesy’ style or tone. You can dial up the formality. You can try and impose a few rules.
But when the client organisation has a distinctive tone of voice, applies a slightly unusual style guide, or has a marketing director that HATES certain words, it’s much easier to ask a human to adhere to those rules than a computer. For me, this is as true of proofreading as much as writing. Read more about editorial style guides.
The stylistic challenge is even more acute when you move into longer copy: a report or white paper. You not only need to maintain a consistent tone, but also build and consolidate an argument or series of findings over several sections. I can definitely imagine a point in the future where AI tools will be able to do this exceptionally well – especially the consistency – but at present, it’s a challenge too far.
The same, incidentally, might apply to ongoing campaigns or programmes, where there is a need for subtle evolution of a repeated message.
Is your copywriter cheating?
Yet, while all of these issues mean I can confidently state that I still see a need for human copywriters, even in the face of the ChatGPT onslaught, I’m certainly alive to the potential of the technology. In fact, I wouldn’t be surprised if, once the initial fun has died down, copywriters become some of the most frequent users of ChatGPT.
It could be a way of producing a first draft at speed, which could then be edited and nuanced to the client style and standards. It could help generate headline or caption ideas, providing a different direction or train of thought which the writer could build on. Or it could be the ideal tool for creating content designed for computers – such as search engine optimised copy.
If that sounds like cheating, or at the very least an opportunity to push back on writing prices, I’d compare it to the “blended” ways of working that many technology-led industries already adopt. Automation and AI are used to accelerate repetitive processes and take on straightforward tasks, all under the watchful eye of a skilled human operator.
Together in electric reams
In short, instead of seeing the advent of ChatGPT as the next stage in the war between man and machine, I believe the best route forward for this generation of writers is to work together.
Let’s just hope the machines see it that way too.