I stumbled across a story detailing a very positive aspect of the potential of AI (which I still believe would be better understood as "MI" for Manufactured Intelligence rather than Artificial. As is traditional, the Universe appears to not care that I am right.). I decided that Google Gemini was the correct author for this post. Everything below this opening paragraph is copied straight from Gemini. This is an incredible use case but I will admit that demonstrations of this level of power and potential are quite likely to gain the attention of bad actors with limited resources. Security must be a primary concern but, we can't and shouldn't halt human progress because a few will abuse the advancements. But we really need to be careful.
The story you're looking for is about Paul Conyngham, an Australian tech entrepreneur and machine learning expert who used a combination of AI tools—most notably AlphaFold and ChatGPT—to develop a personalized cancer vaccine for his rescue dog, Rosie.
While the headlines often say he "cured" his dog, the reality is a bit more nuanced: the treatment successfully shrank her terminal tumors by about 75% and significantly improved her quality of life, though he and his veterinary team are still working on resistant areas of the cancer.
🔗 Key Links to the Story
The Scientist:
ChatGPT and AlphaFold Help Design Personalized Vaccine for Dog with Cancer The Street:
Tech boss uses AI and ChatGPT to make his dog a cancer vaccine The Decoder:
AI consultant uses ChatGPT, AlphaFold, and Grok to find treatment Vet Candy:
A man with no biology degree used ChatGPT... It worked
🛠️ How He Did It
Conyngham didn't just "ask an AI for a cure"; he used his data science background to orchestrate a complex biological workflow:
Genomic Sequencing: He paid $3,000 to have Rosie’s healthy DNA and tumor DNA sequenced at the University of New South Wales (UNSW).
Identifying Mutations: He used ChatGPT to help him analyze the 320GB of raw data to find the specific mutations driving the cancer.
Protein Modeling (AlphaFold): He used AlphaFold to model the 3D structure of the mutated proteins (specifically the c-KIT protein) to see how they differed from healthy ones.
Vaccine Design: Based on these models, he identified "neoantigens"—targets the immune system could recognize—and drafted a formula for an mRNA vaccine.
Expert Collaboration: He took his findings to the UNSW RNA Institute, where real-world scientists verified his work and physically manufactured the vaccine.
⚠️ A Touch of Context
While this is a triumph for "citizen science," experts note a few "don't try this at home" caveats:
Technical Skill: Conyngham has 17 years of experience in machine learning; he wasn't a total novice to data.
The "Cure" Label: One tumor shrank 75%, but others were resistant. It’s an ongoing treatment rather than a final "cure."
Concurrent Therapy: Rosie was also receiving standard immunotherapy, making it hard for scientists to prove exactly how much of the success came from the AI-designed vaccine alone.
No comments:
Post a Comment