Drug scientist: AI speeds up medicine, but the benefits stop at those who can pay
realmeetjames · x · 2026-09-27
The author traveled to Mexico City to interview Dr. Sandra Martínez, a scientist developing obesity and metabolic disease drugs competing with Ozempic, about what AI is actually doing to medicine.
Key takeaways
- A billing problem, not a science problem: AI applied to complex problems like rare diseases only reaches populations that can pay. The science may be finished, but its benefits stop at those who can afford it.
- The language tell: All afternoon, whenever an answer carried real consequences, she switched from English to Spanish. She once said, "I think I couldn't say that in English" — the honest version existed only in the language she wasn't performing in, which is what the film is built around.
- Mexico goes first: In drug development, Mexico accepts pharma protocols and lets its population bear the first-phase experimentation risk, while development happens elsewhere. She described it as the arrangement, not a scandal.
- Faster is the whole product: Asked if things move too fast, she said "Yes. All the time" — citing ChatGPT's post-pandemic spread, feeding AI with open data, ethics debates inside Mexico's institutes, and AI accelerating new protocols and molecules.
The conclusion: the question isn't whether AI speeds up medicine — it obviously does — but who benefits afterward.
More from AGI Musings
- Study: AI access drops people's willingness to say "I don't know" from 44% to 3% — The Decoder · 2026-09-27
- Commenter Claims AI Leaders Use Fear to Push Protectionist Regulation — DavidLinthicum · 2026-09-27
- Blanche Minerva's five-step pipeline shows why "just next-token prediction" no longer fits LLMs — BlancheMinerva · 2026-09-27
- Shopify CEO Tobi Lütke on pruning, AI at Shopify, and whether CEOs will be replaced — yacineMTB · 2026-09-27
- User says Opus 5.5 is the first model to pass their personal Turing test — Charuru · 2026-09-27
- Top mathematician Boaz Barak: hand-crafted proofs are over, embrace AI-era math — mattturck · 2026-09-27