I Asked a Robot to Paint Mona Lisa Holding a Full English — Then Took the Results to Brighton Pier

On Monday morning, full of optimism and supermarket croissants, I typed four words into an AI image generator: "Mona Lisa holding fry-up."

What came back changed me. Not because it was good — although parts of it were disturbingly good — but because it raised a question that has been rattling around my head like a 2p in a cup of tea: if a machine can generate a Renaissance masterpiece holding beans, bacon and a hash brown in 11 seconds, what hope is there for the rest of us? And more pointedly: what hope is there for Terry?

We'll get to Terry. Terry works on Brighton Palace Pier. Terry has opinions about robots.

Week One: The Prompt Diaries

I gave myself seven days and a rule: every prompt had to be something a bloke down the pub would genuinely ask for. No fancy art jargon. Just the good stuff.

Day one. "Mona Lisa holding a Full English." The AI delivered a plate of food so photorealistic I briefly got hungry, and hands so numerous I briefly got frightened. Six fingers on the right. Four on the left. A combined total of ten, which sounds right until you look at where they are. Rating: 7/10 — delicious but haunted.

Day two. "Mona Lisa waiting at a bus stop in the rain, Grimsby." Genuinely moving. She looked like she'd been waiting since 1503, which, to be fair, she has. 9/10.

Day three. "Mona Lisa as a Brummie auntie at a wedding disco." The generator gave her a fascinator the size of a satellite dish and put a glass of Prosecco in her hand. Somewhere in Birmingham, a real auntie was violated by this image. 8/10.

Day four. I asked for "Mona Lisa doing a weeing-in-a-garden gag, dad at barbecue energy." The AI refused, citing policy. A felt-tip and a caravan window have never once refused me anything. Point: humanity.

Day five through seven are a blur of her on a barge, her in a Greggs, her holding a pint of Best outside a pub called The Smirking Cavalier. The best output — her, mid-smirk, on a deck chair in Margate, a 99 flake melting faster than her expression — was, I admit it with a heavy heart, brilliant. 10/10. I would hang it in my hallway.

Enter Terry

Brighton Pier on a drizzly Tuesday is a specific kind of theatre. The caricaturists sit in their little kiosks like lighthouse keepers of comedy, chins out, pens twitching, waiting for stag dos and granddads.

I showed Terry the Margate deck chair picture first. He looked at it for a long time. Then he said the thing I'd been dreading: "That's better than what I'd do."

Then he looked at the six-fingered fry-up and perked up considerably. "But it can't do this," he said, and in ninety seconds flat he'd drawn me with a nose like a canal barge and a chin you could moor at. It was cruel. It was accurate. It was funny in a way the AI outputs weren't, because the joke was aimed at me, personally, with malice and love.

That's the difference, Terry reckons. "The robot can paint her holding a fry-up. It can't decide you look like a fry-up." His colleague two kiosks down, who asked to be identified only as "the da Vinci of the prom," put it more bluntly: "Let it do the backgrounds. I'll do the insults."

So, Verdict?

Here's my honest, slightly philosophical takeaway, delivered as promised with just enough depth to justify sharing this at work: AI can out-produce the caricaturist. It can generate a thousand Monas before Terry's finished his first chin. But it can't sit on a cold pier, clock your ridiculous haircut, decide you look like a sad owl, and charge you a tenner for the privilege of knowing it.

The pun, the insult, the cheeky grin — that's still a human franchise. For now, the machines can make her smile. Only Terry can make her smirk.

Terry for the record asks that any robots reading this stay off his pier. He's not joking. Neither, somehow, is she.