OG’s AI Editorial Guidelines

Well, since I published my first thoughts on AI for creatives last December, the chat around ChatGPT has only got louder. (Other brands of AI overlord available.)

In the last five months, I’ve spent an unhealthy amount of hours following the developments. As OG’s Head of Copy, it seems only right that I dedicate a fair chunk of my brain-space to figuring out: how can we use technology well, to benefit ourselves and our clients? Also: is this gonna’ nick my job or what? 

Hopefully not, on the second point. As for the first, there’s no quick answer. AI-generated content is an emotive, complicated issue. Many creatives are staunchly pro or violently against; exhilarated or terrified. There are ethical, practical, creative issues to grapple with – and a fairly urgent discussion needed around what this means agencies and our clients.  

Here’s the thing: I don’t know the answer. What I do know is that the best path forward is to stay informed, keep learning, and communicate clearly and openly with clients about what we discover along the way. 

To that end, we’ve set up an AI steering committee (myself, our Head of PR Toby Brown, and hotshot strategist Matt Wyatt). We’ll be sharing regular insight with clients and comrades across the industry alike, reflecting on where AI is serving us, and where it’s no match for good old-fashioned people power. 

First on the order of service: we’ve produced the first draft of our AI Editorial guidelines, to provide guardrails for everybody at OG whose job involves writing. Here it is, for full transparency – we’ll check back in regularly as the picture evolves and we make changes.  

Treat AI as a tool for writers, not a replacement. 

ChatGPT is a handy tool for a writer looking to research efficiently, work faster, spark ideas, visualise an idea, or summarise information. However, it is just that – a tool, not a replacement for a real specialist doing that task. 

If you weren’t a skilled writer before ChatGPT, you won’t be one with it. Even dedicated AI content writing services acknowledge that its output requires robust human editing. Everything has to be checked and reviewed by us, just as it always was. 

So: nobody at OG will be using ChatGPT to produce first drafts of copy. Instead, skilled and talented writers/editors will use ChatGPT to make their work even better. 

Tread carefully, with every prompt. 

In the words of ChatGPT itself, generative AI: 

• May occasionally generate incorrect information
• May occasionally produce harmful instructions or biased content
• Has limited knowledge of world and events after 2021

That’s a big part of the reason we won’t be condoning the use of AI to produce full drafts of content. Research is a skill. A good copywriter has to do a lot of research, interrogate their sources and act appropriately. For example, we would generally not cite information from sensational sources or tabloid newspapers such when producing content for clients.  

If you write something purely using AI, you then need to spend several hours checking the claims within it. Otherwise, we risk exposing ourselves and our clients to a variety of risks: making irresponsible or incorrect claims; publishing factual errors; committing unintentional plagiarism, or publishing content that is biased or contains cultural insensitivity. 

For more context on this issue, read the following article. A reader emailed The Guardian, trying to track down an article mentioned as a source by an AI tool. It sounded like something the journalist in question would have written (it was the sort of topic they covered in their writing), but the journalist had no memory of writing this particular article. Why? Because the AI had made it up as a source. 

As another article notes: OpenAI’s own tests showed that GPT-4 could deliberately lie to a human worker (“No, I’m not a robot. I have a vision impairment that makes it hard for me to see the images”) in order to get help solving a captcha test designed to block non-humans.

There is also a longer-term legal issue at play. Some countries are banning AI, global tech leaders have called for its progress to be halted due to ‘profound risks to society and humanity’, and regulators around the world are working swiftly to address the issue. All in all: we need to be careful about what we do. 

(Side note; we must always avoid posting client data, sensitive information or proprietary data into any AI model, an issue that’s already creating a lot of problems.)  

We’ll always ask ourselves: what is this content for? 

AI generates low-cost content, quickly. In a world where many businesses are time-poor and cash poor, that feels like a win. 

But AI tools like ChatGPT can only gather together information that already exists and repackage it in a new format. It can’t be truly original or breathtakingly creative. And I’ve yet to read anything written by AI that was truly engaging, original or interesting, without a hell of a lot of human editing and intervention. 

And if the content we produce isn’t original, engaging or interesting – why would we want to put it out in the world? What is our motivation for producing content that doesn’t add actual value to the reader?

We also need to look to the future. It’s feasible AI-generated content may be penalised by search engines in years to come, so while churning out content using AI might feel like a winner right now, it might actually have a negative impact for client content. 

What next? 

The next key point of our plan is to get everyone trained on prompting effectively, and to develop a key set of use cases for every department of the agency. We’ll check back in about a month’s time to share that, too – along with a toolkit for clients on using AI effectively and safely. Until then: prompt wisely, and do reach out for a chat if you’d like to know more about how OG is using AI. 

I’ll leave you with a quote about the value of human creativity in the face of AI… 

If your view of the world is one in which profit maximisation is the king of virtues, and all things shall be held to the standard of shareholder value, then of course your artistic, imaginative, aesthetic and emotional expressions will be woefully impoverished. We deserve better from the tools we use, the media we consume and the communities we live within, and we will only get what we deserve when we are capable of participating in them fully.