How does AI affect copywriting? Here’s my honest response.
- 20 hours ago
- 10 min read
Updated: 3 hours ago
My life as a freelance copywriter dramatically changed with the launch of ChatGPT in 2022. Here's what’s happened since.
A friend asked me the other week how AI was affecting my daily workload. Naturally, I wanted to kick a lamp-post and scream at the sky, shouting expletives about ChatGPT, Claude and Gemini.
Instead, I asked him if he truly wanted to know how it worked in my day-to-day life. Foolishly, he said yes.
I think it’s a story worth retelling. Why? Because most people have no idea what it’s like to be a copywriter living with AI. If that sounds like I’m assigning some kind of deadly disease status to it, that may not be far off the mark.
First, though, a quick history lesson.
AI and copywriters: the beginning
ChatGPT first surfaced at the tail end of 2022. Like every other copywriter in the world, I started noodling with it to see what it could do. It was pretty impressive. It was also rather worrying.
If a free bit of software could re-create what we freelance copywriters were being paid to do, what did the future hold in store?
These fears partly came true in 2023 when a great number of lower-end copywriters around the world saw their ‘commodity copywriting’ jobs disappear.
Completely replaced by free AI tools, the number of freelancers who disappeared around this time must have been in the tens of thousands. But the tech was still quite new, so larger projects, such as full website copywriting and planning gigs (my speciality), were still landing at my feet.
In fact, 2023 was one of my best years ever as a freelancer. But it was clear that AI was getting better – and that I needed a new strategy.
The shift to AI copy editor
Let me just spell something out before I get too deep into this. Some kinds of copywriting, such as ad copywriting, may always benefit from a human copywriter working alone – or with a designer. When a client wants something truly original, I believe that a human will continue to outperform AI for some years yet. Maybe always.
The thing is, that kind of work is not the day-to-day reality for many freelance copywriters.
It certainly isn’t the case for me. My main area of expertise is website copywriting. What does that mean? My strategy – in brief – is to consider your business goals, USPs, market and competitors, and come up with a plan for the entire site. I’ll make sure it’s logical, reads well, has powerful CTAs, and I’ll even deliver it with rough designs to take to your designer.
It’s a rigorous, multi-step process, and creating a website – even quite a short one – typically takes a week or more. Larger projects that require lots of writing and planning may take over a month.
Most of my clients today recognise that the most suitable copy for this kind of project is not generated by a copywriter working in a darkened room and "being creative". They want something that works.
It's a bit like when you take your car in for servicing at the garage: you don't want the kid at the back who can tune your Kia so it sounds like a Ferrari. You want the bloke in greasy overalls who knows engines inside out and will follow the rules.
With AI assistance, I can make headlines snappier (or at least stress-test the ones I have come up with); I can make sure that I follow the latest best practices; I can shave 20 minutes off searching for the right couple of words by asking AI to fill in the blanks. And I can get a second or third opinion before filing.
I truly believe that for many copy jobs, me plus AI is better than me working alone.
So that’s where we are today. Some of my jobs are still completed without any AI assistance; others are a kind of co-production with me acting as the AI copy overlord.
What may be surprising is that it's not much quicker working with AI – if at all. When I get input from Claude or Gemini, it’s because I believe it will help me deliver the strongest possible copy for the client – not so I can knock it out in 15 minutes.
How AI has affected my workload
Getting back to what I told my friend, here’s a rough paraphrasing of what I said:
Fewer enquiries
I still rank well on Google for ‘website copywriter’ and ‘website copywriter UK’ in particular. I also do nicely for keywords like ‘brochure copywriter’ and a few others.
But because everyone now knows AI can create usable content, fewer people want copy from a copywriter.
Fact.
Worse, Google AI overviews have affected search engine results page (SERP) visibility. Overviews are the things at the top of every search inquiry that you now see before actual business listings.
Google introduced these in 2023. Whereas once upon a time I may have been fourth or fifth on a page, I’m now often right down at the bottom. And the difference this makes is massive. The number of clicks a typical site generates diminishes by around 50% when it falls from first place to second. When you drop further still, they almost cease.
Better work
The upside of this is that I no longer get calls from people with pocket-money budgets. If you’re starting a business, the last thing you feel like spending money on is a copywriter when AI will create your content for free.
When calls do come in – usually from outbound marketing efforts rather than Google search results – they tend to be from larger clients who see value in my career profile/credentials. This leads to bigger jobs, often lasting a few weeks or more. Four or five years ago, many jobs that came in were done and dusted in less than 48 hours.
Shifting patterns
Every time I think I get a measure of AI and its effect on marketing teams across the UK, something changes.
For example, this time last year, I thought my life as a copywriter was over. Work went quiet because Google searches were down, and AI – just as it will continue to – had become significantly better. But then I felt a small shift.
Late 2025, it seemed like marketing departments were buying into global fears that AI-generated copy wasn’t resonating with readers.
Humans were back in.
Fast forward to the summer of 2026, and work was solid again. But it’s a different world. Everybody I speak to in marketing today is using AI – and none of them ask me not to use it during my work. They accept it as part of the process, and some actively encourage it.
So I move with the times.
I know that if I do use AI, the work I deliver will be meticulously developed – by me – and maximised to the best of my abilities. AI does not make copywriting easy, nor does AI stop me writing the actual copy. It's more of an assistant and – oh, God – 'buddy' rolled into one.
Here are five things almost every piece of copy has to go through when a client tells me to use AI.
1. The brief
A client wanted me to create a master page a few weeks ago that could be replicated with small amendments across the company website. They explicitly asked me to use AI to write the copy, but wanted me to make sure it hit the spot.
Creating the AI prompt to get this off the ground took an hour.
I first had to consider what a good page would look like and assess all the elements it should include – this is editorial judgement, conversion strategy and a bit of SEO rolled into one.
I added keyword parameters, info about the reader, and I told AI to check some specific links for relevant information. I also flagged up the brand's tone and explained a list of style points that would minimise corrections later.
2. First draft
It was awful. I know that sounds odd, but when you write for a living, you get a feel for copy the second you read it.
I suspect my client might have been OK with that first draft, but given that I’ve been working with them as a copywriter for quite a while, I knew it was poor. The AI copy took them in the wrong direction, with an overly casual tone that felt bizarre. I explained this to Claude and asked it for a second version.
3. Second draft
It still wasn’t right. I picked up on several points and told AI not to do certain things, setting it off for a third run.
But a quirk had crept in: On draft one, I told AI it had made a mistake about the brand – which AI immediately elevated to the status of ‘interesting fact’. Instead of simply not repeating the mistake, as I intended, AI seemed to think it was a good idea to mention it as a positive.
I’ll give you an example. If you tell AI that your construction company actually doesn’t use electric vehicles, there’s a high chance it may mention “we don’t use electric vehicles” in every future post. Which, of course, no one would actually write.
Another:
You: "Hey, don't mention that we're not ISO 9001 certified – we're actually ISO 14001 certified." AI: "Established in 2004, the company is ISO 14001 certified, though not ISO 9001 certified..."
4. Third draft – accepted
You might think that by now I’d have a fantastic bit of copy.
I didn’t. It still needed at least an hour of editing to make it ‘feel’ right. Here’s the thing: in writing, there is often no definitive right or wrong – but every experienced copywriter has a very clear gut feeling about whether something passes muster or not.
5. Getting a second opinion
The AI model you use will often get stuck in a groove, which is a weakness when it comes to self-correcting or finding a better way. So an additional run through a second and even third AI model flags up issues you may have missed.
Some suggestions I will happily ignore, but others are valid. So more amendments are needed.
6. Fact checking
Everybody’s heard that AI hallucinates – and my God, is this true. AI can become so convinced of its own brilliance that it will state things as gospel – even when pushed. When you go digging, you’ll find some of the things you’ve been told are pure bullshit.
Checking a single article can take anything from 10 minutes to 10 hours, depending on length, complexity and – quite possibly – how moody your AI is.
The final touch is usually proofing (twice) through Grammarly Pro, and making sure nothing in the text matches anything else online.
7. Job done
The page took around four hours to complete. Which isn’t nothing – so why didn’t the client just get me to write this page from scratch? It’s because they knew it would take me twice as long, based on previous work I’d done (and I told them as much).
The AI approach saved them some cash, and – less good for me, but what can I do? – also gave them a repeatable template for future pages.
If you're wondering what this assignment would have looked like if they hadn't asked me to use AI for the writing but knew AI was part of the deal, let me tell you.
It would have been quite different. I'd probably have asked AI to come up with a basic framework, and then written the first draft myself – bringing AI in for light assistance ("Give me 10 ways to say x") and to critique the copy when finished. Steps four to six would remain – to varying degrees.
The ‘everybody is an AI expert now’ conundrum
With my friend’s attention starting to wane (the drool was the giveaway), I tried a new tack. I thought he’d be interested to hear about how AI has now become so embedded in marketing workflows that everybody is an expert (he wasn't).
For me, this expertise is a real problem.
Sometimes I’ll submit work that I’ve spent a significant amount of time on, and a contact will then run it through their own AI. And because every AI model has its own nuances, it may come back with a completely new take on proceedings. Which the marketing person may embrace... and feed back to me.
It’s only happened a few times, but it’s going to happen more and more.
Suddenly, I'm back at square one. All my work, all my decades of experience, wiped out in a second, on the whim of a different AI.
In consolation, I sometimes imagine how wide of the mark most pieces of business copy (not shiny, award-winning creative ads) delivered today would be if AI hadn’t already given them the once-over.
Those comments coming back from the marketing team might not be, “AI said why don’t you try and do it this way?” – they’d more likely be: “I ran this through AI, and you scored 5/10.”
If you read that and inferred, “Well, this guy must think his own work sucks,” you’re wrong.
What’s happened is that the bar has been raised. Today, everyone expects AI-level precision and Grammarly-proofed perfection. And in a real-life, time-pressed business setting, most humans can’t match it.
The future of AI and where copywriters fit in
So where does this end?
Here’s my take on it. I think within the next 12 to 18 months, every business with 30 or more employees will have fully embraced AI’s potential. They will manage most of their writing needs in-house, using people like me as a final, high-level editor.
I think they’ll still value human judgement – and that is ultimately what people will be paying for, rather than writing skills. It will all be about validation before pressing print.
And that’s pretty scary. Because with this comes the expectation that a job that once took five days should be completed by AI in minutes, then given final oversight by an experienced writer/editor in a couple of hours. That would be fine if I could charge £500 an hour, but – yeah, fat chance.
Instead, I think rates will stay the same – as will the volume of work. For a freelance copywriter who once completed 20 jobs a year at a couple of grand a pop, there’ll still be 20 jobs a year… at £300 each.
Which is unsustainable. I may soon have more earning potential as a street urchin.
Prodding my friend back to life, I quickly summarised that AI has completely upended the copywriting world; that yes, I was pissed off about it; of course I was trying to adapt as best I could; and – hey – I might be completely wrong about this and six months down the line marketing bods will be flooding to hire experienced copywriters like it was 1999.
He smiled and said, “Wow.”
My sentiments entirely.
Footnote: AI scored this post 8/10. There's life in the old dog yet. Written by Mike Peake, actual human being UK website copywriter + freelance copywriting services (emails, ads, brochures, etc.).
T: +44 (0)208 133 4306
Need help with your copy? Get in touch for a free quote (and a chat about AI if you're interested). I can work with AI on or off – the choice is yours.
