An AI cancer vaccine actually worked. A man used ChatGPT, Grok, and AlphaFold to build a personalized mRNA vaccine for his dying dog's cancer, and the tumor shrank by half. This week on They Might Be Self-Aware, Hunter and Daniel tear apart the story, debate Claude as your post-op physician, and propose a formal intelligence rating system for kitchen appliances.

Daniel had face surgery and immediately pasted his medical notes into Claude like it owes him a consultation. Turns out AI medical advice is surprisingly useful for the 80% of questions that aren't life-or-death, if you prompt it right (Daniel did not prompt it right). Hunter explains the sycophancy problem and how to get honest answers from an LLM instead of digital hand-holding.

Then the main event: an Australian man's dog was dying of cancer. He took ChatGPT and Grok down a rabbit hole that led to AlphaFold, a university professor, and a custom mRNA cancer vaccine, the kind of personalized medicine that could eventually win a Nobel Prize. The tumor halved. The dog came back to life. Daniel says AI drug discovery is going to change everything. Hunter says just ask the AI.
Andrej Karpathy released "auto research": AI agents that autonomously optimize machine learning models. He gave each agent one hour to beat his hand-tuned results on a small GPT. They beat him by 11%. The hosts get into hyperparameter tuning vs. real architecture changes, and whether spawning an AI researcher with a one-hour lifespan is an ethics problem or just Tuesday.

Finally: Philips put a "conversational virtual assistant" in a coffee maker. It's a questionnaire. Daniel is furious. This somehow leads to the invention of standardized AI intelligence levels: Level 0 (the Philips coffee maker) through Level 7 (you'll have to subscribe to find out).

⏱️ CHAPTERS

0:00 Gary Calls from a Raccoon Payphone
1:21 Daniel Fed His Surgery Notes to Claude
7:24 AI Cancer Vaccine Saved a Dying Dog
13:56 Karpathy's Auto Research Beat His Own Brain
22:38 The Fake AI Coffee Maker

⚡ Listen now & get self-aware before your tools do.

🎧 Listen on Spotify: https://open.spotify.com/show/3EcvzkWDRFwnmIXoh7S4Mb?si=3d0f8920382649cc
🍎 Subscribe on Apple Podcasts: https://podcasts.apple.com/us/podcast/they-might-be-self-aware/id1730993297
▶️ Subscribe on YouTube: https://www.youtube.com/channel/UCy9DopLlG7IbOqV-WD25jcw?sub_confirmation=1

📢 Engage

Dog has cancer? Just ask the AI. What's the wildest thing you've actually asked an AI for help with? Drop it in the comments.

New here? Subscribe for twice-weekly AI chaos.

🧠 They Might Be Self-Aware — but are we?

#AI #AICancerVaccine #TMBSA