Use the robots. Keep the weird.
AI can make us faster. Let’s not let it make us beige.
Chantal Palmer | Head of Brand and Creative Strategy
Recently, Hayley was interviewed as part of a research project looking at the impact AI is having on teams. They talked about trust, psychological safety, leadership and team dynamics, basically, what changes when AI becomes part of the way a business works?
Oooh, lovely. Right up our street!
Having a little nosy into the transcript, I realised a thing. Although we’ve spent the last couple of years testing AI, building AI into our workflows and generally being gobsmacked at all the things it can do, the actual conversations we were having weren’t really about the technology but about knowing when to use AI and, perhaps way more importantly, when not to.
Begin at the beginning
The researcher mentioned that from the organisations she’d spoken to, Keystone seemed to be further along than many in the way we’d integrated AI into the business. Which was a surprise to me. I thought that we were probably way behind (imposter syndrome much) but as it turns out, we’re flying the flag!
However, we didn’t start with a masterplan for AI transformation. It started with me poking it with a stick to see what happened. There was no AI expert to explain it all, just me and my insatiable appetite for science fiction, realising the future had arrived in my lifetime!
To give that excitement a focus, Hayley and I went to an Agency Hackers event called The Robots Are Coming, which introduced us to just how enormous the potential of AI was. Exciting? Yes. Overwhelming? Also yes.
It would have been very easy to disappear down an AI-shaped rabbit hole researching everything without actually doing anything. Equally, “EVERYONE GO AND PLAY WITH AI!” didn’t seem like a particularly sensible business strategy either. Especially when you’re dealing with client information, intellectual property, strategy and ideas. We needed to learn to walk before sprinting enthusiastically into the robot-filled distance! At some point, curiosity needs guardrails.
We invested in a paid business version of ChatGPT, created an AI policy and agreed what information could go in, what shouldn’t and where responsibility sat. We started mapping where AI could genuinely help across the business rather than wedging it into everything because it was new and shiny.
AI became a team sport
I also hadn’t realised until reading Hayley’s interview that introducing AI from the bottom-ish-up rather than the top-down might have made us different from other organisations too. Our genuine curiosity drove adoption, we learnt together. Someone found something useful and shared it. Then someone else did. A prompt worked brilliantly? Shared. ChatGPT produced a really useful way of analysing something? Show everyone. Found a limitation? Talk about that too. For a while it was basically an AI love-in.
And yes, we gave our AI a name. Of course we did. She’s Nova. Champagne SuperNova if she’s in trouble. Her pronouns are she/her/overlord. Is anthropomorphising an artificial intelligence entirely rational? Probably not. But giving her a name made it easier to talk about how what we were asking AI to do and what role she should play.
Spaghetti brain.
I’ve learnt that my favourite thing about AI is getting through the messy middle. My brain is not naturally linear. There are thoughts. Lots of them (I’m not sure if you noticed : ) Bits of information. Questions. Connections. Something somebody said three weeks ago. An idea that may or may not be relevant. Another idea. Something shiny. Back to the first idea.
And this is where AI has been transformative. She can structure, research, analyse, challenge, remember, summarise and pull all the threads together.
I can give Nova the contents of my brain and ask her to help me corral it into something I can actually work with. What am I missing? Where is this argument weak? What assumption am I making? What would someone who disagreed with me say? Is there another way to look at this?
The more thoughts the merrier. She is a brilliant sparring partner in that process. But I don’t ask her to come up with the concept. Absolutely not. I think that’s what us humans are best at. Put people in a room to come up with a thing and they wander off topic. Someone says something ridiculous. That reminds someone of something apparently unrelated. Someone says, “Hang on…” And there it is. The big idea. We’re quite good at spotting the interesting turning three miles earlier that everybody else ignored. I’m not ready to hand that bit over.
The risk of beige
Competent is becoming super easy. AI can produce perfectly decent copy, imagery, research and ideas at extraordinary speed. That democratises access to skills in a way that I think is genuinely amazing. But if you’re asking AI to create something based on everything that already exists, unsurprisingly, it’s pretty good at producing something that looks like everything that already exists. Competent becomes much less distinctive.
Have we become so seduced (isn’t that a lovely word!) by how quickly it can get us to a decent answer that we stop asking whether decent was actually the thing we were aiming for?
Our value lies in what we refuse to hand over.
AI hasn’t made us less busy. Sadly, none of us at Keystone are now reclining on a chaise longue on Thursday afternoons being fed peeled grapes because Nova has done all the work. What she HAS done is remove some of the friction and give us more space for the bits of our thinking that is valuable enough to protect.
I don’t want AI deciding which thought has potential before we’ve had a chance to kick it about. I don’t want efficiency to become the objective when the actual objective was originality. And I definitely don’t want us producing perfectly polished beige stuff simply because we can produce more of it, faster.
To stand out, organisations will have to figure out when AI can help them steer the boat better and when they still need a slightly weird human being to notice the treasure that everyone else sailed past.
