With more channels for communicating than ever before, there’s a higher demand for content and copy than we’ve ever known. With the recent explosion of generative AI (artificial intelligence) that’s readily available to all, copywriters might be seeing their clients turn to AI tools to create articles, blogs, and social media ads or they might be using AI to support their own writing.
But is this at a cost to credibility of organisations, not just copywriters’ careers?
As one of those copywriters, it’s hard not to be biased but I’ve tried to put on my neutral hat as I discuss the point: ‘Should we be using generative AI in marketing copy, more specifically in content writing for life sciences organisations?’. As you read on, you may notice my neutral hat does slip a little – apologies for this, I am only human after all.
The subject of using AI in content creation can be a divisive one. For many, it saves precious time in research and in the writing itself, but is this at the detriment to originality of thought? We can go into schools and see this happening. I’ve spoken to teachers who set homework as summarising an article or writing the main messages as bullet points and, of course, many students give this task to AI – job done! Does this mean we are deskilling our youth? That’s perhaps a discussion for another time but a point I wanted to raise.
According to Microsoft Copilot, which (helpfully or unhelpfully depending on your viewpoint) now pops up on Word when you open a new document, “Generative AI, powered by advanced machine learning algorithms, is rapidly redefining the way marketing copy is created and delivered.” And while this may be true, at least for some companies, it’s worth thinking about the capabilities of AI, where the information is coming from and if it’s credible. AI-written text can read so well that it can sound as though a point is true even when it’s not.
What is generative AI?
In IBM’s words, “Generative AI refers to deep-learning models that can generate high-quality text, images, and other content based on the data they were trained on.” Large language models (LLMs) are the subset of generative AI that can decipher language and can generate text.
The key here is ‘based on the data they were trained on’. From everything I’ve read it seems that AI, at present, does not have the functionality to determine whether data is true or not. And, let’s be honest, AI output is not always high quality either – I’d argue the point that even if something is well written, if it’s total tripe then that does rather bring the quality of it into question. Though on the other hand, On the Ning Nang Nong and The Squirdle are two of my favourite poems and they’re utter nonsense. But I don’t think Spike Milligan was trying to sell a product or service with those poems – he was an entertainer, a natural linguist, but not a marketer so his words could run off the tongue beautifully and not need to mean anything beyond what the reader inferred. Okay, so I’m off topic but if you don’t know these poems and you do know a child or can access your own inner child, these poems give a wonderful few moments of down time.
Right, back to AI!
Useful AI tools are ChatGPT, DeepSeek, Perplexity, Claude and Gemini, to name a few that have free initial access though I couldn’t manage to get into DeepSeek on my iMac but that could easily just be a me thing. Grammarly has been using AI for years, just not talking about it till more recently “AI was central to our product long before it was a buzz word,” but it also uses your data (unless you opt out) to train its language models. This is where you need to read the small print; some AI tools automatically opt you in and some don’t. This might be one of the reasons why we’re finding that more and more life sciences companies are requesting that AI NOT be used in any copy or design creation.
Here’s how a few different AI tools claim they can help with the creation of marketing copy in life sciences after I posed the question: Can you answer in one sentence, how can [name of relevant generative AI] help with life sciences marketing copy?
- ChatGPT
“ChatGPT can help with life sciences marketing copy by translating complex scientific concepts into clear, compliant, and compelling messages tailored for patients, professionals, or stakeholders.” - Perplexity
“Perplexity can help with life sciences marketing copy by generating accurate, research-backed content with real-time citations, enabling marketers to create compelling, trustworthy messaging tailored to scientific audiences.” - Claude
“Claude can help with life sciences marketing copy by crafting scientifically accurate, compliant, and engaging content that effectively communicates complex medical concepts to various stakeholders while maintaining appropriate regulatory standards.” - Gemini
“Gemini can assist with life sciences marketing copy by generating scientifically accurate drafts, tailoring complex information for diverse audiences, optimizing for SEO, and streamlining content creation workflows.”
But where’s the creativity? The strategic thinking? The story telling? Where are the new ideas? Prompt AI to tell you a story and it will reel off a story. We ask it a question and it will answer, but how many people are asking it the same question; how can we be sure AI isn’t giving them the same answer? Do we want our marketing copy to be a duplicate of someone else’s? Not only does this negate any originality of thought (please, let’s not all turn into robots), but the duplication won’t do any good for your SEO either – Google hates duplicate content, so using it is likely to harm your website’s ranking performance in organic Google search listings.
Should we use AI in marketing copywriting?
When it comes to content creation, if we want our work to be credible, to be accurate, to be genuine opinion when it’s an opinion piece we’re writing, to be genuine thought when it’s a thought-leadership piece, we must harness the human brain. We can’t simply rely on AI. I’m a big believer in using AI for research for certain topics – planning where to go on a mini break for example, or, more relevantly, giving ideas for planning a marketing campaign or researching different points for an article.
But should we use that information without checking it first? No! In the life sciences sector, we generally love stats, we’re programmed to love them – beautiful numbers that tell us information in a clear way. So, when you put a question into Google or into ChatGPT and some stats come up that can substantiate the point you’re making it can be tempting to grab those numbers and go with them – there’s your evidence, fabulous! But we must check our sources and if that reference isn’t the original source of the stats, what is and was their research credible? Because one thing AI can grab onto and spit out as an answer is duplication – if enough people quote the wrong statistics, they pop up front and centre and can be written in a way that they seem to be true.
From my own experience of source-checking articles created by others, I’ve noticed that where stats have been taken from some AI tools, there seems to be many layers to get through to discover the original source of some of these figures and that some of these sources are not credible at all. And AI itself can create errors. A nonsensical output of AI is called an AI hallucination: this is a “phenomenon where in a large language model (LLM)… perceives patterns or objects that are non-existent or imperceptible to human observers, creating outputs that are nonsensical or altogether inaccurate.”
A recent study published in Nature warns us of “AI model collapse.” This Oxford University-based study found that: “training AI models using data generated by previous models leads to long-term learning problems. The models degrade in quality and can ultimately fail when trained on recursively generated data, i.e. when they ingest data that they produced themselves. The study attributes this to the build-up of minor errors and misconceptions over time, inherent to machine learning models. These are then reproduced by subsequently trained models, which in turn add slight errors of their own. Over time, these can cause the models to collapse”.
This might not affect your choice at the moment, but I believe it’s worth knowing as it can impact the integrity of the data AI tools generate.
I’m not a hater of AI. Am I an advocate? Maybe, probably. I know a lot of people who are excited about the capabilities of AI (and also a number of folk who are worried about their job security) and for content and copy writers it can be very useful in assisting to transcribe a conversation or to pull out specific points from an interview. But if you are using AI to grammar check your work, polish your draft, or create a draft from a recorded transcription, ensure you’ve checked the security and privacy policies of the AI tool. Some AI tools use your text to train their LLMs by default. Some, by default do not, but, as I’ve already mentioned, do read the small print.
Should life sciences organisations use AI in copywriting?
Many organisations, especially those in the life sciences sector are requesting that AI is not used in the creation of their assets. This seems like a sensible approach to maintain confidentiality and originality of work as well as compliance, with AI not necessarily being able to interpret regulatory standards.
Another consideration when using generative AI to create your marketing copy is that AI cannot have empathy, it’s not sensitive to the needs of others and won’t have the nuances of language or the abilities to perceive the wants and needs of your specific target audience.
McKinsey says of generative AI models: “sometimes the information they generate is just plain wrong…Organizations that rely on generative AI models should reckon with reputational and legal risks involved in unintentionally publishing biased, offensive, or copyrighted content.”
Generative AI tool, Perplexity, tells us: “Generative AI is best used as a starting point or support tool for copywriting rather than a standalone solution. It excels in efficiency but requires human input to achieve high-quality, engaging content tailored to specific audiences.”
Generative AI can be an extremely useful tool to increase efficiency of output (it can rewrite a campaign email for different platforms, for example) and aid with research. But if you are using generative AI to help you write your copy, ensure confidentiality and privacy restrictions are in place for sensitive information, take time over your prompts, always read your content before it goes out and ask yourself: does it sound like something your organisation would write? Is it in your company voice? Will it resonate with your target audience? Check the sources are credible before you publish.
In my opinion, generative AI can be a wonderful tool to aid research, to pull out points from an article or conversation and can even be useful in creating a draft of an article if you’re in a rush and happy to risk using duplicated content, but always consider your end goal and take whatever AI throws at you with a pinch of salt. I believe AI is an amazing tool and can save you time, but the human brain is so much more amazing – if you do choose to use AI-generated marketing copy, I’d advise you to always include human input and oversight. And, if you’re using AI to write thought-leadership pieces, you might want to think again.
*****
Karen Noble of Noble Creative Writing is a valued member of Lean Creative Marketing’s network. At Lean, our default stance is to avoid using AI to create marketing assets without explicit consent of, or at the request of, a specific client. If you’d like to find out more about our services and how we can help you to produce high-quality copy and design assets to help your organisation stand out from the crowd, get in touch.
