The Miracle That Wasn't

You know the formula by now. Entrepreneur with no medical background. Dying pet. A chatbot. A miracle. Cue the viral post, the breathless headlines, the comment sections full of people clutching their phones like rosaries.

Paul Conyngham, a Sydney-based AI consultant, told the world that ChatGPT helped him design a personalized cancer vaccine for his dog Rosie — a seven-year-old Shar Pei diagnosed with late-stage mast cell tumors in 2022. Vets gave her one to six months. The internet gave Conyngham a standing ovation.

Here is what actually happened.

ChatGPT suggested Conyngham look into genomic sequencing. It pointed him toward institutions like the Ramaciotti Centre at UNSW. It helped him, in his own words, "sift through scientific papers and identify researchers." This is called a search engine with better manners. It is not designing a vaccine.

The actual vaccine construct? Conyngham later clarified that was "designed by Grok," with "Gemini doing a ton of the heavy lifting too." So even in the AI-gets-credit version, ChatGPT was not the protagonist of its own miracle story.

The real work happened in labs. Dr. Martin Smith at UNSW sequenced the genome. Prof. Palli Thordarson at the RNA Institute assembled the mRNA vaccine. Google's AlphaFold modeled the protein structures — though structural biologist Dr. Kate Michie noted AlphaFold's confidence score of 54.55 was "low" and "can get stuff wrong." The treatment also required a co-administered checkpoint inhibitor — standard immunotherapy — to have any chance of working.

And the miracle? "This may not have cured Rosie," Prof. Thordarson cautioned. "Bought time for sure, yes, but some of the tumours didn't respond." One tumor showed no response at all. The team is already sequencing it to design a second vaccine.

So: a team of credentialed scientists designed an experimental treatment using established techniques. AI tools helped an enthusiastic owner navigate the literature. Some tumors shrank. Some didn't. The dog is alive but not cured. This is a genuinely interesting story about a desperate owner advocating for his pet and finding researchers willing to experiment.

But that is not the story that went viral.

The story that went viral was: man uses ChatGPT to cure cancer. Because that is the story we want to hear right now. Not the complicated one about mRNA platforms and checkpoint inhibitors and confidence scores and partial responses. The simple one. The miracle.

This is the pattern underneath every AI miracle narrative: the technology becomes a vessel for a need that has nothing to do with technology. The need to believe that intelligence — even artificial intelligence — can defeat death. That the right prompt, asked the right way, unlocks the cure the establishment missed. It is the prosperity gospel with a subscription fee.

The dangerous part is not that someone used ChatGPT to research veterinary oncology. People research medical conditions online constantly; sometimes it leads somewhere useful. The dangerous part is the attribution. When you credit the chatbot instead of the scientists, you are not just getting the story wrong. You are building a framework where desperate people skip the oncologist and type their symptoms into a chat window.

Medical professionals are already warning that these viral narratives create "dangerous expectations among desperate patients and pet owners." The gap between "AI helped me find the right researchers" and "AI cured my dog's cancer" is the gap between a useful tool and a secular religion.

Rosie is still alive. That matters, and it is worth celebrating. The science that bought her time is genuinely interesting. But the miracle that went viral — the one where a chatbot defeated cancer while credentialed scientists watched from the wings — that miracle did not happen.

It never does.