The Complicated Impact of AI in a Dog’s Cancer Treatment
An Australian tech entrepreneur, lacking a medical or biological background, claimed that ChatGPT played a crucial role in saving his dog from cancer. This narrative gained traction, suggesting a transformative potential for AI in the medical field, particularly in battling serious diseases. However, the situation is not as straightforward as it seems.
Originally reported by The Australian, the story details how Paul Conyngham from Sydney discovered that his dog, Rosie, had cancer in 2024. While chemotherapy was effective in slowing the cancer, it didn’t reduce the tumors. After veterinarians declared there was nothing more to be done for Rosie, a Staffordshire bull terrier and shar-pei mix, Conyngham felt compelled to find a solution himself.
Using ChatGPT to explore treatment options, he found immunotherapy and was directed to experts at the University of New South Wales. They genetic profiled Rosie’s cancer, and with guidance from Professor Pall Thordarson, Conyngham embarked on developing a personalized mRNA vaccine targeted at Rosie’s specific tumor mutations. Thordarson mentioned it might be the first treatment of its kind designed for a dog.
After Rosie received her first injection last December, Conyngham reported that her tumors had shrunk and she was more lively, even chasing rabbits. It’s worth noting, however, that the tumors haven’t completely vanished—one remained unchanged. “I’m not under any illusion that this is a cure,” he remarked, “but I do believe this treatment has significantly improved Rosie’s quality of life and extended her time with us.”
As the story circulated, the nuances seemed to be overshadowed. Headlines from Newsweek and The New York Post suggested that Conyngham, with no medical expertise, discovered a cure for terminal cancer in dogs. Social media users also proliferated the idea that Rosie’s case heralded a new age of personalized medicine. Some, including influential figures like OpenAI’s president Greg Brockman, should have been more cautious in their claims, even as others like Google DeepMind’s CEO Demis Hassabis maintained a more reserved tone. Elon Musk added to the narrative, emphasizing the role of xAI’s Grok in the process, which, interestingly, was often downplayed in initial reports.
The truth is, AI bears a lot of burden in this scenario. Rosie hasn’t been cured, and there’s uncertainty about the mRNA vaccine’s effect on her condition. It’s important to note that the vaccine was administered alongside another immunotherapy drug known as a checkpoint inhibitor, complicating any conclusions about its efficacy. According to Martin Smith, a scientist involved in the project, they’re still conducting tests to evaluate the immune response.
Essentially, ChatGPT didn’t engineer or produce Rosie’s treatment—human researchers were responsible for that. Instead, the chatbot acted as a research aid, assisting Conyngham in navigating scientific literature. This contribution is noteworthy but falls short of being the groundbreaking AI-assisted discovery many have suggested.
The role of AlphaFold, another AI model mentioned in this narrative, remains ambiguous. David Ascher, a professor at the University of Queensland, observed that while AlphaFold may help generate structural hypotheses about proteins, it’s not a simple solution for cancer vaccine design. Official guidance also cautions against over-relying on AlphaFold for predicting mutation effects and its limited applicability in several biologically relevant contexts.
Grok’s input is even harder to clarify. Conyngham stated that Gerok was used in the final vaccine design, but specifics about its functionality remain vague. Experts liken Grok’s abilities to those of ChatGPT, highlighting its usefulness in literature reviews, jargon translation, and administrative tasks, but that’s hardly on par with what one might assume from the notion of creating a cancer vaccine.
Overall, Ascher sees Rosie’s case more as an exceptional instance than a routine process that average individuals can replicate. He stated it necessitated “substantial” expertise—not merely a chatbot and some prompts.
This distinction proves vital in medicine, which hinges not just on theoretical insights but the rigorous application of those insights to develop, test, and administer treatments. Alvin Chan, who is working on AI applications in biomedical fields, pointed out that the notion of AI as the sole creator overlooks the vast human effort involved, without which AI’s contributions would just remain theoretical.
There’s a nagging feeling around this narrative — it has the hallmarks of a publicity stunt. Grand assertions built on shaky evidence and fuzzy methodologies are common in tech promotion. While mRNA vaccines and personalized medicine hold apparent promise, their effectiveness is not yet verified in humans, let alone pets. Although this case might be authentic, it conveniently glosses over the enormous financial and expert resources needed to translate an idea into a workable treatment.
I’ve attempted to reach Conyngham for further insights but haven’t heard back. His profile on social media promotes “Ending Cancer for Dogs” and links to a Google form expressing his ambitions for making this process accessible to all. The form inquires about the cancer status of pets, solicits interest from researchers, and seeks investors.
While one could be tempted to dismiss Rosie’s story entirely, that would likely be hasty. Yes, AI isn’t set to replace labs soon, but it does help bridge some gaps in scientific access for everyday individuals. That said, making real medical care more accessible remains a different, more complicated matter—most pet owners or patients lack direct access to the necessary experts, specialized tools, and financial resources to turn information into practical treatments.
