Like most people, I’ve been wrestling with what generative artificial intelligence means and, more accurately, means for my future.
Officially my job title is executive creative director. Will it soon be executive prompt director? (Yes, I know that’s basically what it is anyway.) Or will I be replaced entirely?
A couple of weeks ago, I found myself turning to my now old friend, ChatGPT.
It was my turn to write the end-of-week agency-wide email. A blast of positivity, global updates and often tipsy Friday evening stream of consciousness.
The aim is to keep us all connected and end the working week on an up. Everyone feeds the writer a few titbits of news and then they craft it all together into a cheery missive.
But Hannah, our wonderful account director, was out of the office. I knew that if her news wasn’t included, she’d be missed. I wondered whether ChatGPT could write it for her.
A few prompts went in: account director, executive producer, Parisian, loves film. Some lines of relatively chatty text came back. I stuck it in the email under her name and didn’t tell a soul.
Then I asked people to spot whose update had been written by AI. They spotted it immediately. Within seconds. Of course they did.
The process of doing this I found to be reassuring. I was a little scared of generative AI beforehand, but the more I tried it the more I felt that it wasn’t putting me out of a job any time soon.
All it did was highlight what clients are really paying us for: original thinking.
It reminded me that good writing is good thinking, and to write a good prompt you need to be a good thinker. I spent a lot of time honing my prompt. By the time I’d got it to spew out something vaguely approximating Hannah, I might as well have written it myself.
But that wasn’t the point. The point was I needed to think about what I wanted to convey. The meaning and the intent. It sounds obvious but ChatGPT can’t think. It’s merely an execution tool.
Good writers need to be good thinkers, which means the good ones won’t be out of a job any time soon.
And then there is tone of voice. The whole exercise was a dramatic reminder of why tone of voice is everything.
When I looked at all the different updates in the email from all the different staff members, I could see so clearly that everyone has a unique, utterly idiosyncratic and odd way of writing.
It’s so individual, it’s like our thumb prints. Hannah is half-French, half-English, working and living in two countries, with a background in war-based documentary-making and a fervent dislike of flowery language. Put that in your chatbot.
Of course she was going to be spotted as a fake immediately. How do you mimic that? Yes, you could feed it all the words Hannah has ever written, but it still won’t know that Hannah unknowingly says the phrase "a small splash in the sea" instead of "a drop in the ocean."
In many ways, this could be the start of a new age of creativity. Because the more work that gets churned out by ChatGPT the more samey everything will feel.
Generative AI after all is just an auto-predict on steroids. It can only write based on what’s been done before. Which means it’s North Star is stuff it recognises and looks familiar.
That’s what we are all in a battle against, right?
As my old ECD used to say to me, “agencies are just a bunch of smart people in a room thinking." This remains the truth of it.
Brands still need to find something unique, and interesting, to say. And then say it in a way that is distinct to them. That requires thinking and we can help them do this. In fact this role may have just become more valuable.
There could now be an even greater premium on originality. Oh how I hope so.
Coda: the tech has changed, ignore all the above. This article has not aged well.
Jim de Zoete is executive creative director at Across the Pond.