Both personally and professionally, one of the most common topics of conversation for copywriters these days is generative AI.
Can AI really replicate ‘human’ writing styles and produce professional quality content? How can you use ChatGPT to write a press release or blog? Do you worry about the impact AI will have on your business?
These are all questions that copywriters need to take seriously. While generative AI might still be a ‘dirty word’ in some circles, the reality is that many businesses and their content teams are turning to it to assist with their output – from ideation and planning through to writing and sense checking copy.
And it will not be going away. The ChatGPTs, Claudes and Jaspers of this world will only continue to evolve and become more sophisticated. The question left for us copywriters is – how can and should we respond?
Whatever that response, I believe it should be based on some real knowledge and understanding of how these models work. That is why I’ve spent some time getting used to them.
Here, I will attempt to present some of the things I’ve learned along the way, and how those learnings have shaped my stance on what role AI can (or should) play in the world of copywriting.
Benefits of AI for copywriting
Let’s start by looking at some of the advantages.
If generative AI tools are used properly (more on this later), then there are some pretty clear benefits to be had. These include:
- Saving time: ChatGPT and other generative AI models can produce text at speed, meaning the only time you need to commit is in formulating a brief. The impact of this grows as the tools learn every time they are briefed.
- Gaining analytical insight: One of the most impactful use cases I have found is AI’s ability to dissecting and analyse large volumes of information. For example, it can spot patterns in surveys and datasets that you might take hours to spot or miss altogether. Alongside this, it can also break down long transcripts to extract key talking points.
- Spotting non-subjective errors: AI tools can be excellent proofreaders and act as a second set of eyes on your work.
- Overcoming writer’s block: We’ve all been there. A topic is put in front of you, and you cannot get off first base. Or you run into a roadblock halfway through writing an article. In these situations, a quick prod into an AI engine can set the wheels in motion. For instance, you could ask it to suggest a few broad angles or ways to approach a theme, or how to go about a sticky section of an article that you are unsure how to articulate.
Limitations of AI for copywriting
ESG performance can be converted into a score, known as an ESG rating.
That’s the positive case for leveraging AI tools in our line of work. It’s quite compelling when taken at face value, but it is important to stress that these gains can be easily outweighed by a minefield of drawbacks.
Some of these drawbacks are avoidable (again, if the tool is used properly), while others are less so.
Based on my experiences so far, these are the main pitfalls that need to be avoided:
- Inserting unfounded assumptions into copy: Untruths can creep into AI generated text. These can include incorrect and unwarranted assumptions, as well as ‘facts’ which are unsupported and cannot be substantiated when the AI is queried about them.
- Formulaic writing style: To a trained eye, ‘bot’ copy can be easy to spot. There are patterns in the writing to look out for, such as overuse of particular sentence styles, use of punctuation and unusual individual word choices.
- Over-reliance on AI for factual (and up to date) information: This is arguably the most serious problem. Generative AI tools should not be leant on to provide detailed, substantiated factual arguments without some serious human oversight. Furthermore, the information they draw on is not live, which opens up the possibility of being provided with outdated content.
- Loss of human creativity and critical thinking: Because of the apparent ease and speed at which AI tools can generate copy, the possibility of falling into lazy habits is real. If these models are overused and taken as gospel, there is a real danger it could damage our capacity to innovate and think creatively, as well as question the content being presented to us.
You only get out what you put in
When asked about how to use ChatGPT and other AI tools, my default response is ‘the more you put in, the more you will get out’.
This is based entirely on the experiences I have had. The likes of ChatGPT and Claude will produce better outcomes if you put in the effort to brief them properly.
Depending on what you are tasking it to produce, a ‘good brief’ will include details on structure, style and target audience, as well as reliable information to base the content on. Simply asking it to create a press release about a new product with a few cut and paste details on said product will not be enough.
This then begs the question – if I am spending a lot of time feeding a substantive brief into the AI, then why not just write the content myself? How much time will I actually save by doing this?
Human oversight is a must
What’s more, even if your brief is as thorough as it can, it is also critical to sense check everything that the AI gives you in return. This, once again, adds to the time you need to commit to the task.
Whether it’s editing for style to ensure it hits the right tone, checking the use of facts and assumptions, or simply seeing if quotes are prescribed properly, there are many boxes to tick when it comes to producing a successful piece of content.
AI models are far from perfect, and anybody leaning on them to assist with copywriting tasks will quickly discover this for themselves. What you envisage, and what the AI generates, can be far apart even if you put in the miles to brief it thoroughly.
What is our take on using AI?
At this stage, there are more questions than answers, and more concerns than surefire wins.
We certainly see potential in using AI tools to aid with traditionally laborious processes such as analysing large documents and transcriptions where privacy concerns aren’t an issue, as well as providing an initial burst of inspiration to help with unfamiliar or sticky topics.
However, AI is no substitute for human-produced content of the type we typically provide for clients. Using robots to get close to the quality and originality of what we can create ourselves, especially for longer form content such as thought leader bylines, blogs and white papers, would arguably take longer than writing it from scratch on our own.
This is not to dismiss AI or take the moral high ground in any way – on the contrary, we do not believe the Luddite position is a sustainable one for the copywriting trade to take.
As the models evolve, we need to keep up to speed with how they work and adjust our position accordingly. Setting out a stall firmly against AI, or ignoring it altogether, is not something we can afford to do.
Do you want to find out more about our copywriting and editing services? Contact us to discuss how we can support your next project.
