I was at my allotment on Easter break when Ru, Chicken’s Co-Founder, sent me a podcast to listen to, he said it was fascinating… I was intrigued.
It was an interview with Simon Willison about AI. He’s a programmer, co-creator of Django. Ru went to university with him and he used to freelance at Torchbox before me, and was very much remembered in a lot of conversations whilst I was there. So I knew he was a clever chap so it was probably worth listening to even if I really didn’t understand how AI could be fascinating.
We do a little with AI at Chicken: built a Drupal module that works much better with it, use it for a few boring tasks, occasionally to check my writing or untangle a complicated problem. Nothing that had ever made me think: oh, this is something.
I wasn't sure I wanted to listen, but I pressed play and carried on digging. It's potato planting time. It’s the the only time I actually dig, preferring the no-dig method. Some of what was said went over my head. Some of it sounded worrying.
At one point he said something that made me put down my fork and stop digging.
The funniest thing about AI, he said, is that we've built the most expensive, power-hungry computers in human history, and if you ask them to draw a pelican on a bicycle, it looks like a five-year-old did it. He said it like it was the best thing about the technology. Like the absurdity was the point.
Asking the LLMs to draw a pelican on a bike is Simon’s benchmark for how good the thing is. Playing to assess quality.
Something started to unlock, it wasn’t scary. It was like playing with a 5 year old.
I have a memory from childhood. I must have been about ten, staying at my best friend Lynne's house in winter. We were in her back garden making witch's potions.
The way any good budding witch does:
- Gather leaves, flowers, bits of bark, feathers and sneaky bits of our hair
- Stir into a tortoise-shaped bucket.
Then it was tea time. Then it rained. Then it froze.
The next morning, we found it.
It was not how we left it, it was ruined.
We squeezed the ice block out and polished it between our hands, melting the surface until it went clear, until we could see the leaves suspended inside. We brought it inside to show Lynne’s mum, who promptly sent us back to the garden for dripping all over the place!
I remember thinking it was the most beautiful thing I'd ever made.
We hadn't really made it.
We'd made the conditions for it.
The rain and the cold did the rest.
Here's the thing: what we stumbled into by accident, as two children pretending to be witches, is now a 4 step craft. Grownups do it deliberately. Fancy ice for cocktails. Adults had to rediscover and formalise what we'd just done by playing.
There's actual research on this. Play in adults isn't frivolous — it's the state in which we learn fastest, make unexpected connections, and recover from the relentless pressure of having to know what we're doing before we start. Most working life is designed to eliminate it. Outputs are expected. Mess is inefficiency. We stopped making witch's potion a long time ago.
James Victore — whose book Feck Perfuction I've read cover to cover, and whose Creative Warrior course I've done — makes this argument more loudly than anyone. The gist: you don't make good work by waiting until you can do it perfectly. You make it by doing it badly, repeatedly, until something emerges. The bad versions aren't failures. They're the process.
And then along comes AI.
I've been doing this with this very site. The brief I gave was something like: accessible, elegant, nature-based, Erik Spiekermann and William Morris rolled into a 2026 vibe. Previously I would have tried to prescribe every little thing I wanted the website to be but the machine took it, and we made something I'd been trying to imagine for a long time.
I want to be clear that I'm not naive about it. These tools are expensive to run, energy-hungry, and the hype is exhausting. There are real questions about what we're building and who it serves. I hold all of that.
And yet I have never felt more like a ten-year-old pressing flowers into a tortoise bucket whilst doing my job.
Because the stakes are low.
Because you can ask something stupid and see what happens.
Because sometimes you get a sloppy mud pie, and sometimes you get something that stops you — something you didn't plan or expect, that you find yourself looking at the next morning thinking: how did we make that?
That's not productivity. That's play.
What Willison was laughing about — the pelican, the five-year-old drawing style — is the thing that makes this interesting to me. The machine is not infallible. It makes weird mistakes and strange leaps and sometimes it's just wrong. And somehow that makes it a place where you're allowed to experiment, because you can't fully predict what you'll get. The only way to find out is to try.
I'll admit I had some fear about not having a job in the future. But I also see friends who haven't had work for the last year or so because Drupal isn't in favour right now. It's tough out there. The thing is, it's circular — it always has been. And when the badly-built AI sites start needing fixing, that's when people who actually speak the language will matter again. A marketeer can prompt an LLM to build a site. They won't notice when the pelican is missing its bill.
Chicken exists so we can live more freely, be creative, and play. So that's what we're going to do with AI. Play.
Over the last four years there have been plenty of times we've worked flat out and the outcome felt very small. We want to make decisions from a calmer place — more grounded, more rooted to what we actually care about, as a business and as people. AI gives us a chance to embrace work differently. So we're leaning in.
Lynne and I haven’t really spoken since secondary school. The ice cube lives only in memory now. But the lesson it holds is still good: you don't always know what you're making while you're making it.
Sometimes the rain fills the bucket. Sometimes you come back in the morning and something beautiful has happened.
Go make some witch's potions.