How I use AI
AI, and specifically LLMs like ChatGPT, Claude and Gemini have really taken off in the last few years.
ChatGPT was just starting to gain a bit of traction in 2022 as I went on maternity leave. Google launched AI summaries (which were hilariously bad in the beginning) as I came back to work in 2023/24. And since then, our reliance on them has really snowballed.
So I think it’s important that, especially given my craft, I lay out my own rules on how I will and won’t use AI when working for writing and editing clients.
I’ll be honest, I don’t love AI
I can’t lie about this: I am not the biggest fan of AI, and that might put you off working with me. I have many reasons for this that I’m more than happy to talk about (although you might regret asking – I’m passionate and ranty)
But I’m also a realist. AI has many advantages that can help make us all more efficient. But I also have boundaries with it and I hope those are evident in the rules I’ve laid out below. I believe that if someone wants to work with me as a writer or editor, it’s because they value my skills as a human, and I hold that value in the highest regard.
My cross-client, cardinal rule
This is the one rule that applies across all of my clients, whatever the task:
If you ask me not to use AI, for a specific thing or at all in the work I do for you, I won’t use it. Simple as that, end of.
How I will and won’t use AI
The following rules assume we haven’t had a separate conversation about AI, in which any preferences you have about AI use would supersede my default positions as explained here.
Writing tasks
I won’t run any content you give me through any AI model or LLM without your express written permission. This includes the brief, any brand or tone of voice guidelines you share with me, and any other examples of your brand’s writing shared for reference. Additionally, I will rely on you to give me this permission proactively; I will not ask for it. I don’t want you to think it’s something I want to do if I’m allowed to. In fact, even if you say it’s OK for me to feed your content to AI, I still probably won’t.
I may use AI in the research phase of my work. I do this because if a topic is complicated or I’m not familiar with it, it could take me a couple of days to learn enough that I can write about it, and more importantly, sound like an authority on it. AI helps me speed that up, keeping costs down for my clients. Here are some (but not an exhaustive list) of the ways I might use AI to aid my research:
To get a general overview of a topic or issue
To distill the key takeaways from a long document, like the transcript of a parliamentary debate or a publicly published report
To understand what others have already written about this topic or issue (competitor research), the gaps that remain (that I could cover for you), and so on.
I will always fact-check AI output. Even though AI can help me research thoroughly but quickly, I will still fact-check the information I’m given with credible sources.
I categorically do not use AI to write for me. When you get a piece of written content back from me, you can rest assured that I wrote it using sentences formed in my own brain, and typed out using my own fingers. I don’t even use AI assistants like Grammarly when I’m writing (I tolerate spellcheck).
I will allow you to run the content I provide you with through AI models or LLMs if you wish. The content I create for you is your intellectual property, not mine (unless we have something different agreed in writing, which would supersede this). You may use AI and LLMs as part of your own processes – for example to score how well a piece of writing satisfies your tone of voice or brand guidelines (or both) – and that’s fine. If you want me to use AI to carry out checks like this before submitting work back to you, then that would come under my first rule.
Editing tasks
I will not run your content or manuscript, in its entirety or even in contextless parts, through ANY AI model or LLM. Even if you give me permission, I still probably won’t.
I may sometimes use AI to ask a question. For example, if I need to look up a grammar rule to make sure your syntax is correct, or check the accuracy of a date, source, or other detail included in your writing, I may well go off the Google AI summary response.
I will manually fact-check AI output. Lord knows AI is still far from infallible and makes a lot of mistakes, so even if it’s given me an answer to a question, unless it was reaffirming something I thought I knew already, I will double-check it with a credible source just to make sure.
I do not use AI writing/editing assistants like Grammarly. While Grammarly and others provide fantastic resources and hubs for language and grammar knowledge, I don’t use their AI assistant plugins while editing.
Admin
I don’t use AI. Whether I’m preparing a scope of work document, a contract, generating an invoice, or designing my website and branding, I either do it myself using creative tools I’ve learned to use or templates available online (for free or to purchase), or I work with another trusted human professional.
If you have any questions about how I will and won’t use AI in my daily working practice that aren’t covered here, please feel free to get in touch.

