“We hope to reimagine the drug discovery process with the dream of one day solving all disease.”
That was the line. Delivered with a flat face. By Google DeepMind boss Demis Hassabis. At Google I/O.
Big statement. Huge, if it’s true.
But it’s not magic.
It’s called Gemini for Science. A bundle of experimental AI tools for researchers. That’s it. No crystal ball.
I’m usually hard on health AI here in Optimizer. Hassabis deserved context he didn’t get.
Science comms are broken. We need stuff simple enough for my grandma to understand without spreading fake news. Hard. The researchers in that room got it. AI shrinks the timeline for medical breakthroughs.
The rest of us?
We heard “Gemini cures cancer.”
We hear that AI fixes everything.
It doesn’t work that way.
AI is old news in labs. Wearables? That’s machine learning. Detecting stuff without sticking needles in you? Algorithms. Generative AI is newish though. Promising, yes. I talk to clinical folks often. They see it. AI sped up COVID-19 vaccine dev. Massive win for humanity.
But the catch? Ethics. Bias. Privacy. Who gets access?
Those problems didn’t go away.
Hassabis mentioned two projects.
AlphaFold figures out protein structures. Proteins do everything in your body. If you know their shape, you might unlock cancer treatments. Scientists just found 1,700 new proteins that could help. Finding these used to take years. AlphaFold does it faster. Case study: malaria vaccines. Parkinson’s clues. LDL cholesterol targets.
AlphaGenome predicts DNA mutations. Might explain why diseases start. But Google admitted limits. In Nature, no less. It’s not validated for personal genomes. Misses cell-specific patterns. Nuance gets lost on everyone but scientists.
So why say “solve all disease” on stage?
He wasn’t talking to you.
Or me.
But the quote leaked out. Spread fast.
These tools won’t erase cancer in five years. Or ten. Maybe twenty. Maybe longer. For science? That’s fast. For your sick grandma?
Slow as molasses.
Hassabis had forty other AI announcements. He had to move. Soundbites travel far. And our relationship with AI health? It’s been messy. Regurgitated data. Hallucinations. Annoying prompts.
It’s human to conflate lab tools with consumer apps. But don’t.
Then comes the comparison trap.
Remember RFK Jr.? The health secretary said AI could make the FDA irrelevant. Different context. Same buzzwords. Same audience confusion. Does Google agree? No. But people think so.
Kennedy wants to skip trials. Skip animal tests. AI doesn’t erase rigor. It assists. Tools need humans. Rigor cannot be skipped willy-nilly or otherwise.
Context is king.
It always dies first in social clips. Short videos kill nuance.
My wellness grifter rulebook says: Juxtapose a true fact with a misleading claim. Is that a crime here? Probably not. Google and Apple do real clinical work. But telephone games distort reality. Declining literacy. Low attention. I plug holes where I can. That’s my job.
Sciencewashing is rampant now.
Buzzwords buy legitimacy. Erase doubt. Silicon Valley bros attend peptide parties. Worship Bryan Johnson. Optimize their sleep. Hack their bio.
AI cures all → Track your metrics → Buy these pills → Defeat death.
It’s a slippery slope.
Will AI eventually cure diseases?
Maybe. One day.
But the road won’t be clean. Twenty years is a long time in politics. In society. In culture. Clinical research shifts. I’m not betting on Hassabis’s optimism today.
Wait and see.
