Generative AI: without a human hand, heart and mind, it’s mute
Artificial intelligence is best paired with human intelligence, creativity, curiosity and craft.
When it comes to writing, it needs us.
And that’s why we’re not out shouting ‘the robots are coming and they’re going to take our jobs!’. Being fearful of AI would be like us clinging to typewriters in a world of tablets. Instead we’re embracing it, and showing our clients how they can too.
But first, let’s tackle that word ‘artificial’
Generative AI tools are a marvel. Bard sits blinking on my desktop every day. But it’s mainly my research partner, thesaurus, and outside of work, meal planner. If it was applying for a job as a junior copywriter, I wouldn’t hire it.
(Side note: PR agency Bottled Imagination hired and fired their AI assistant for being boring, self-critical and fabricating information.)
Why wouldn’t I hire it? Because it’s not a great writer. It’s fast. But not great. It lacks imagination and the ability to break the writing rules to get to something excitingly unfamiliar. The writing feels over-worked. It’s…artificial.
Here’s what Bard (now known as Gemini), wrote as the opening paragraph when I gave it the brief for this article.
Friends, let's crack open a cold vérité on the AI cocktail du jour: generative language models. The scene's intoxicating, swirling with promises of automated copywriting, chatbots with Molière's wit, and marketing campaigns churned out like assembly-line sonnets. Ah, the allure. But for us here at [Agency Name], we're pouring a different blend. One spiked with a secret ingredient: the human touch.
A bit ick.
But, pair AI with a human being and things get exciting.
AI + a heartbeat
Here are a few ways you can bring generative AI into your writing and editing process (with a little help from a certain group of language consultants).
1. Train a generative AI tool to write in, or check for, your tone of voice
We recommend the tools that learn by being fed writing samples. Because you can train it on the best examples of your tone of voice in action. Written by talented, human writers. So it’s picking up all the right knowledge and habits. (We can help you create that data set, train the tool, test and refine it.)
If you’re creating a tool to tell you whether something is on brand or not, the same thinking applies. Looking to your tone of voice guidelines to help you train that tool is a start, but not the end. Get your brand language gurus involved, and don’t miss an opportunity to get that tool looking for what’s inclusive language and what’s not.
2. Chatbots
You could turn your brand language guidelines into a bot. I can well-imagine a world where everyone has an AI writing coach sitting on their computer, offering advice on the words people put on the page. Like Clippy of the 1990s, but good. And less annoying. The benefit? You’ll build skills in your team, save time on rounds of feedback and rewrites, and stay on-brand, more of the time.
Those chatbots need some top-quality training so that they’re genuinely useful, accurate and staying true to brand. Create those bots without input from the creators, guardians and champions of your tone of voice, and you’ll be getting magnolia writing that blends in with everything else and everyone else.
3. Prompt writing as a life skill
Good prompts = good responses. The prompt ‘rewrite this more clearly’ will get you somewhere. But ‘please cut this email down by 50% and use language that a 9-year-old would understand’ will give you something you can really work with.
Training people to become better prompt writers is good for them, and good for business.
We see prompt writing a bit like touch typing. Touch typing saves people time and cognitive effort. Being able to write good prompts with fluidity, and at speed, will be a skill that becomes more and more valuable over time.
Training your people to speak to AI will benefit them in and outside of work.
Where to start with generative AI overall
If you’re in a big business, chances are there are chatbots already working away inside your organisation. Their job is to save people time and put information at their fingertips. PwC’s ChatPwC serves up tax and regulatory info to staff. Saving them countless hours.
Brands big and small, have chatbots talking to their customers. For many businesses that’s the main point of contact. A chatbot can be your brand ambassador – the perfect, polished embodiment of how your brand looks, sounds and behaves. Or, they can be a strange, sterile, add-on to a brand’s presence. Creating a weird disconnect for customers.
Our question would be: can those bots be better?
And more specifically….
- Are the responses clear enough to be understood by all of the intended audience?
- Does it use inclusive language?
- Is it using your brand’s tone of voice?
- Does it have a personality and a name, that builds a connection with people?
- Can it deal with off-topic questions and cope in a crisis?
Answer no to any of those and it means you’ve got an opportunity there, to make more of what you have.
My big prediction is…
Human-crafted content will be something brands come to pay a premium for.
There’s science that says as humans, we prefer things that have been created by humans, over computers. In a 2009 study, researchers found that the pleasure centres in participants’ brains lit up when looking at human-created art vs computer-generated.
This HBR article hypothesises that in a world of ‘synthetic creative outputs’, the uniqueness of human creativity will be seen as more valuable.
It’s already happening elsewhere in the world of branding. Like Nikon’s ‘Natural intelligence’ campaign railing against AI-generated images. And Apple’s 2023 holiday campaign that showcased stop-motion animation. They clearly show the presence of human craft.
So why not in the world of language?
Final question – did any of this article come from AI?
Yes. The word ‘alchemy’ in the title. Thank you Gemini.
What next?
Visit our dedicated AI page to learn more about AI at The Writer.
