The ConsultaChem View on AI-Use in Content Creation
With the increasing use of AI in everyday life, especially in the preparation of written content, we’ve put together a short piece about the use of AI by the ConsultaChem team.
On a recent prospective client call we were asked if we used AI to produce written content for our clients.
Our answer was emphatically, “no”.
Which then led to the question, “why not?”. While at the time, I gave an answer that suited the situation, this issue is actually something that warrants further discussion. Especially if our clients – both prospective and current – are thinking the same.
The ConsultaChem standard is that we don’t use AI in the preparation of pieces for our clients, whether it’s preparation of a piece from scratch, adaptation of materials for other outputs, or compiling a ‘top 10 takeaways list’ or summary materials derived from an academic paper. While on the one hand we could be called technophobic, on the other, this was a conscious decision, with our reasons behind it outlined below.
However, before we discuss opting out of AI-use in client-facing work, we should make it clear that our stance on AI for our work does not mean that we are anti-AI. There is a time and a place; in some situations it can be useful and there are some good applications of AI that could genuinely be life-changing. However, we don’t think that these include writing technical content.
Reason #1: We need to understand our clients’ product or service
Whether we’re preparing an academic paper, a textbook, or a longform blog post for a website, our priority is understanding the product or service that our client is selling. While AI can be used to give an overview, our experience is that this overview is superficial at best, and mis-leading at worst.
While it may take 10 seconds to generate that initial output, it then takes us many more hours unpicking it and rewriting it. And what’s the point in that? Our clients are paying us for our knowledge and understanding, in fact they’re paying us because we have in-depth subject-specific knowledge. And if the immediate team working on the piece doesn’t have that knowledge, we have an extensive network that we can turn to. This approach is proven to work.
A few years ago we were working with a client to write a whitepaper that outlined a use-case for a very technical product. It wasn’t my primary area, but I understood enough to pull together a first draft. However, I know I was out of my depth – I needed the help of another professional who had years of experience in the field. So, the network was invoked and their input was used. And while it cost us money (and reduced the profit), what actually happened was that an error in the underlying data was uncovered, so our client re-collected and re-analysed their data. We then prepared a piece that was scientifically accurate and contained a better model of the product our client was selling. Was this over-and-above the scope of works? Yes. Did it make us feel better to know we helped a client avoid publishing marketing materials that were inaccurate? Absolutely. It was our network that helped us, and it’s our network that’s an added value (included in the price) when you work with us.
Reason #2: Technical scientists can see through AI slop
While the AI machine is still very much a ‘shiny new thing’, it feels like the tide is turning and the generation of content for content’s sake is starting to become a real sticking point. While in some industries you may get away with it (although the jury is out on that), in technical industries where a substantial percentage of workers are high-level scientists with Masters- or PhD-level degrees, or you’re working in a highly regulated industry such as healthcare, AI-generated marketing copy really needs to be closely monitored to make sure it’s absolutely on-point. If a reader suspects AI, or if the writing isn’t accurate, there’s a strong chance that they won’t engage – or worse, you may get a call from the regulatory body. In these cases it doesn’t matter how good your product is. If the writing is misleading or incorrect, then it’s on your head.
Good technical writing requires a good technical understanding of the field. And good technical understanding only comes through hard work and graft. However, being an expert in an area doesn’t automatically make you a good writer. Writing well is all about framing the context, explaining the context clearly, and then leading the reader through in a logical fashion.
Reason #3: Client IP
While it may be tempting for some to run a quick client idea through Claude or ChatGPT, as Arthur Weasley says in Harry Potter and the Chamber of Secrets “Never trust anything that can think for itself if you can't see where it keeps its brain”. While we’re not suggesting AI is harbouring parts of Lord Voldemort’s soul, there is a real risk through use of an AI program that IP may end up in places that we didn’t intend. We consider this seriously. With most clients we sign an NDA and in many cases clients give us ‘too much information’, which we then pare back and frame into the output they need. As far as we’re concerned, that means we need to be trusted. Feeding material into an LLM is a no-go.
Reason #4: Human-generated content performs better
Since the mass-adoption and deployment of AI to generate more content for content’s sake, in some cases there is a back-lash. While short-term there may be gain in website traffic and hits, longer-term the trend is that website hits fall – which is apparently linked to Google’s own algorithms. In fact, Google says that the best performing content:
Gives a unique point of view, with real decisions and real context.
Creates content that is helpful, reliable and puts the reader first.
Focuses on what readers want, without laying it on thick.
Reason #5: We love learning
It may be a cliché – and it definitely doesn’t help pay the bills! – but we write because we love learning and supporting others in science, rather than for the money. Call it a hang-over from academia, where it was definitely for love rather than money….
When we’re working with an R+D team to create a marketing piece, we need to quickly understand complex issues and then ‘translate’ them into customer-facing content. And, as we said under Reason #1, we need to understand our client’s service or offering. It’s absolutely crucial to write about it well and get into the shoes of a customer who may rightfully ask, “why should I care”.
Something else that’s important is the ‘translation’ of the science, a skill honed from years as an academic – you try making FMO theory accessible! – and through putting yourself in the place of the reader. Context is key, as is knowing when a diagram to explain a concept is necessary.
Reason #6: The environmental impact
While up to this point the reasons have been knowledge-based, the elephant in the room is the environment and the impact that the general use of AI has on the wider-world. According to the United Nations Environment Assembly (UNEP), AI datacentres require massive amounts of water, produce electronic waste, rely on critical minerals and rare earth metals, and use massive amounts of electricity. While us keeping our AI use to a minimum, or not using it at all where we can, is unlikely to stop the degradation of the planet, as Tesco says, ‘Every Little Helps’.
ConsultaChem: Real humans writing about science
If you’d like to work with real humans who care about your product or service, then get in contact today. We promise we’ll work with you to understand your needs, delivering high-quality materials on-time and within budget.