This is both a cautionary tale and a tale of curiosity.
What this isn’t, is a debate on the rights or wrongs of generative Artificial Intelligence (AI). There’s been A LOT written about that and I don’t feel qualified to talk about it. Instead this is a narrow aperture of using AI and seeing what it can do.
You’ll notice the photo above of two creatures. On the left is a weird, slightly scary Gothic dog-thing, wearing a twisted superhero outfit. On the right, is a pretty cute puppy, also wearing a red and yellow hero suit. Canine features are discernible: button-brown eyes, floppy ears, paws.
Chris Branch, Founder and Marketing Director of Seedily, shared this image at the start of his presentation at the DO lectures (read here about DO if you missed that post). He used the two photos to illustrate how fast AI is evolving.
While he covered AI across all media, he particularly focused on ChatGPT (for text) and Midjourney (for images).
”This is a decentralised form of creativity,” Chris said. “Your idea can be a single idea and 30 seconds later you have a way to represent that one idea in many ways.”
Reluctantly on the journey
Last year, Midjourney, an AI program and platform, made headlines when their free trial version was shut down because of ”extraordinary demand and trial abuse” said founder David Holz. I totally missed it, but deep fake photos of Pope Francis decked in a white puffer jacket and sporting a diamond chain went viral.
In the words of BBC journalist, Alex Hughes, Midjourney, like other AI art programs “isn’t without its controversies”. However, continues Hughes, it leads in painting and Gothic/sci-fi inspired artwork, as well as regular updates “in techniques and advancements in its training database.”
This is where the puppy comes in.
Getting the experts in early
During his talk Chris Branch said that the first 20,000 users of Midjourney were art directors from around the world – all part of the ‘training database’. He explained that the same prompt – along the lines of ‘puppy dressed as a superhero’ – was used in June 2022 and a year later, in June 2023.
As you can see. The results are exponentially different.
A year after testing and learning, harnessing global expert feedback, refining the technology, the AI-generated puppy image is transformed.
No wonder there are likely to be new jobs coming, titled ‘Prompt engineer’.
Putting ChatGPT to work
About a year ago, when OpenAI ChatGPT3 launched, I typed in, ‘what is a regenerative business?’ In about 10 seconds it spat out an okay answer: vague, wordy, lacking any specificity. Still, I was impressed how it described an abstract concept. After that I switched off. As a writer, the ethical minefield of AI felt too deep to navigate.
But I’m a pragmatist.
So, a few weeks ago I thought it would be fun to experiment. I asked my niece Imogen Scobie, who supports me with social, to attend an online course. Lazy Discipline ChatGPT Edition is run by David Hieatt/DO lectures. Imogen got a first class degree in philosophy and is fascinated by the ethics question of AI.
Here’s the 3-step process Imogen took.
Step 1: She created what David calls ‘Anchor Notes’ – a document giving the relevant context and prompt. In this case, says Imogen, “I started with a description of who Claire is and what she wants her LinkedIn posts to achieve. Then I wrote a piece of content in Claire’s writing style.”
Step 2: Then she input that content PLUS an example of somebody else’s writing style. ‘In the style of…’ gives ChatGPT added context and a specific example to work with.
Step 3: Finally, she gave ChatGPT a prompt like: “Write three versions of this in the style of Claire so that it is more engaging and appealing for LinkedIn”. Imogen could then choose which versions she liked. “I also asked ChatGPT to alter the style, tone down the adjectives or make it more appealing to LinkedIn’s algorithm.”.
While the results were pretty good, we both felt the posts were overwritten. (And btw, we haven’t published them yet. We’ll let you know if we do!)
Imogen’s 3 top takeaways:
- Use ChatGPT as a tool to make the quality of your copy better. Not as a quick fix. You might not save time, but you will be more efficient at coming up with better ideas…
- “The quality of your questions determines the quality of your life”. Success in using Chat GPT comes down to the context you provide and the prompt you ask it. If you ask ChatGPT an unclear question, you will get a vague response.
- It’s you versus you. Get better at being you, instead of trying to be like everyone else! At the end of the day, ChatGPT cannot write better than you can.
Imogen and I would love to know how you are navigating AI.
Are you running towards it, or veering away? Are you being honest when you use it (i.e. quite a few people have shared how they write their posts or emails with it).
Is it a way to become more creative? If you’re experimenting, what are you discovering?
Hi, I’m Claire. Through my business Wordstruck we help companies bring their sustainability strategy to life. As the Founder of Regenerative Storytelling, we’re helping leaders do more for their people, their community and the planet. I publish regular content about storytelling, regenerative leadership and reframing how to address our rapidly heating world. To see more of my content, please sign up – and join the conversation by sharing a comment below.