Where were you when AI discovered a potential new treatment for cancer? I was at my desk, writing this story.
The news, announced by Google this week, October 15, is in a way delivering on a promise that AI would someday solve the world’s biggest problems. It’s the kind of pie-in-the-sky statement that tends to get under people’s skin. AI is not a one-size-fits-all solution. It’s not even one thing. AI is many different generative models handling all sorts of tasks. What they tend to have in common is their size and their ability to solve problems based on that size.
For most people, though, AI or chatbots like ChatGPT, Copilot, Gemini, Claude, and others are best used to boil down complex ideas, create images based on text prompts, and put Stephan Hawking in an F1 race car. In general, AI is not solving the hard problems.
Until, I guess, it is.
AI scientist
New research coming out of a collaboration between Google and Yale University claims that a new 27-billion-parameter foundation Gemma model (C2S-Scale 27B) designed to understand human cells, developed “a novel hypothesis about cancer cellular behavior,” which was then confirmed by the team. The discovery could lead to new cancer treatments.
At the heart of the discovery is a problem that, up to now, no human cancer researcher could solve: how to make “cold” or hidden cancer cells show themselves. The task put before the AI model was creating a drug that could act as a “conditional amplifier.”
While smaller models failed on this task, the giant C2S-Scale 27B model did not, with it looking at 4,000 drugs and predicting which ones would boost antigen presentation. The model identified both drugs that are known to possess these capabilities and “surprising hits” for drugs they did not know could possess this capability.
The paper notes that the researchers then validated the hypothesis with real-world tests in which they tested a combination of interferon and the identified drug (silmitasertib). As predicted by the Gemma AI model, the dosage increased antigen presentation and made the “cold” tumor more visible.
This could be big
I know. That’s a lot, and I gave you the Cliff Notes version.
But think of it this way: many late-stage cancers are not diagnosed, including prostate and breast cancer, because of cold tumors, in which there aren’t enough T-cells present for a diagnosis until the cancer is far advanced and harder to treat. This breakthrough could lead to much earlier diagnosis, and that could save lives.
All that makes this discovery exciting, but that’s not necessarily my point. It’s this moment where the true promise of AI is, if not realized, glimpsed.
Conversations about the pros and cons of AI are pretty common these days, with many people sharing an uncomfortable mix of excitement and fear. AI is a fun and useful tool for quickly ideating, summarizing, and even creating, but it’s also a massive disruptor.
Jobs are changing because of AI (some are disappearing). Professionals in every walk of life use it casually and with real purpose. Doctors might quickly turn to, say, Gemini, for more ideas on a diagnosis. Artists will rough out an idea. Filmmakers might make a low-resolution movie of a scene. There are endless possibilities and just as many questions about where this is all going.
However, the idea that AI could be something more than our assistant and plaything has always been part of the narrative. It’s just that we never saw any evidence of it, say, helping solve world hunger, the climate crisis, or international conflict.
Now, though, we have proof, perhaps, that it has at least the potential to do big, important things that matter to humanity.
That’s a moonshot moment if I ever saw one. I hope there’s more to come.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.
