top of page
Writer's pictureChris Silberston

AI copywriting

Updated: Feb 16


Robot working at a computer desk, surrounded by other human workers

We’re all hearing about generative AI. We’re told that job seekers are using it to write cover letters. Students are using it to write essays. Marketers are using it to write emails and social posts. So as content creators, how worried should we be? And how can we know when to use AI? And when to avoid it?


It used to be when you told someone you’re a copywriter, they either had a blank look on their face (is that something to do with stealing music?) or they were interested to find out the sorts of projects you worked on.


Today, a common response is, “oh so you’re going to be looking for a new job soon then?”


300 million jobs at risk from automation, according to Goldman Sachs

Everyone knows that AI is out to steal jobs. Coders, journalists, lawyers, financial advisors, designers – they’ll all need to retrain as care home staff, dance choreographers, and tree surgeons.


Or will they?


AI is not a direct replacement for real people

AI is getting scarily clever. It can do many tasks far better than people ever could. But artificial intelligence isn’t the same as human intelligence. We can take advantage of the differences to find ways to work with AI – and keep our jobs while we’re at it.


Generative AI like ChatGPT is certainly impressive for a piece of software. And low-skilled content creators should be worried about losing work. But the kind of work that ChatGPT can currently replace is the kind of work that was barely a step above the auto-fill text when you write an email. Talented individuals can think of generative AI in the same way as that auto-fill text – a tool to save time but not to replace the truly creative thinking that only humans can achieve.


Making the most of our human intelligence and creativity is how the coders, journalists, lawyers, financial advisors, designers – and yes, copywriters – of the world can add value.

So how do we use generative AI in a constructive way? First, lets think about what we can do that AI can’t.


The difference between generative AI and human intelligence


Humans draw on varied experiences

We’ve all heard stories of child prodigies that studied Maths at Cambridge at age 8. Or can play every Prokofiev piano concerto after hearing them one time in the womb. This kind of intelligence is impressive, but not what you need when you want to write a business strategy or a novel.


In many ways, AI is like a super intelligent child prodigy. It’s extremely good at certain tasks. But it doesn’t have the worldly experience to make unlikely connections or understand things through intuition.


Humans ask illogical questions

Some of the greatest feats of human accomplishments were the result of mistakes. Or chance encounters. Or someone asking, “what if I put mentos in my coke?”


AI is completely logical. It can imitate creativity and divergent thinking. But it can’t actually achieve it (yet).


Humans can guess what other people are thinking

Until we all have a direct neural connection to ChatGPT, it can’t know how its responses affect our emotions.


It can try emotional techniques – writing about something scarce or rare might make us feel the fear of missing out. But it can’t read its writing back to itself and try to get in the mind of its audience.


Humans are sceptical

There have been plenty of examples of AI-generated lies that were confidently presented as facts. AI can’t tell the difference between true and false, other than being told X is true and Y is false.


Humans can change their minds. We can be critical. We can ask hard questions. We can look at a report from 2005 about the state of the web and think "hmm, maybe I should find something more recent."


If we couldn’t we’d all be scientologists by now.


Humans can be disobedient

If you ask AI to write a headline to generate as many clicks as possible, the result is probably going to be clickbait. It will analyse blog posts about persuasive headlines and just copy some of the techniques and words it comes across.


But if you ask a human writer for a headline to generate as many clicks as possible, a good one should ask “why?” Clicks for the sake of clicks is a pointless goal.


Humans want personal connection

When chess Grandmaster Garry Kasparov lost to IBM’s Deep Blue, it would be easy to give up on the game altogether. Why bother playing something you’ll never be best at? No human can beat AI at chess today.


But chess lives on. As Kasparov put it, “the way Deep Blue played offered no input in the mysteries of human intelligence.”


It might be better than us in many ways, but it’s a long way from imitating the thousands of little signs of humanity that we give away when we communicate.


Humans can multitask

AI systems (at the moment anyway), are specialised. They can chat. Or create images. Or drive a car. Or recognise faces. Each system can potentially do its job far better than a human could. But humans can do all of these and more.


Inspiring thoughts often come from the connections we make between different areas of our lives. A piece of art might influence a business decision, for example.


What does this all mean for your copy?

Some forms of content don’t need these very human qualities to make them successful. These are the obvious candidates for a workflow that includes AI. Others need a lot more thought and are best left to real people.


Promising types of content for AI to write

  • Bulk blog posts or social posts that would have been farmed out to content mills in the past

  • Short pieces of content marketing (basic whitepapers, for example) that don’t need unique insights or perspectives

  • Simple event communications like reminders or invitations

  • Web pages about common products or services

  • Follow-up comms in a sales journey where the only objective is to remind prospects you exist

  • Editing existing content with a simple goal in mind (less words for example)


Complex content not to trust AI with

  • Any content where a Google search can’t provide a quick answer

  • Strategic or brand work

  • Anything that draws on a variety of sources of information like interviews or market research

  • Creative work

  • New concepts or trial projects where there’s no existing format to copy

  • Thought leadership pieces where credibility and interesting perspectives are important


Other uses for AI to integrate into your workflow

  • Speeding up research that would normally take a lot of internet searches

  • Summarising large amounts of information

  • Translating other languages or hard-to-understand technical information

  • Getting some basic content on the page to get you started

  • Generating lists


Watch this space

All of this comes with a caveat. It’s just a matter of time until AI moves beyond us in every kind of intelligence – 2040 is the most frequent prediction.


But when that happens, it won’t matter if you’re a mathematician or a lion tamer, we’ll all be in trouble. AI will no longer be a tool, but a coworker. Or perhaps our evil overlord.


TL;DR. How to use AI for copywriting

Based on the quality of copy we’ve seen to date, we wouldn’t trust AI to actually write anything you want to send out to the world. If you want to sound authentic and engaging, you still need a human involved in the copywriting process.


But AI can be used to save you time as part of your workflow. Especially for simple tasks that in the past you’d give to the intern.


You could ask a chatbot to research businesses in a specific geographic area, for example. Or get it to translate or summarise source information. It could also be useful as a starting point for some basic pieces of content – like simple editing tasks or basic, informative communications.


Here’s another way to think of it. Anywhere your readers would expect something to be written by a person, don’t use AI-generated copy!


Need human assistance?

If generative AI isn't generating results, we're always ready to help. Get in touch!

Commenti


bottom of page