The pursuit of a cancer cure goes back thousands of years. Some of the earliest known research can be traced to ancient Egypt, where Imhotep, the physician and architect for King Djoser, described a human tumor on papyrus around 2600 BC.
Currently, an increasing number of tech leaders are extolling AI as the key to solving the medical mystery that has confounded physicians for thousands of years. This is what president Ruth Porat said last October. And this is why Anthropic CEO Dario Amodei coined the term “,” which reflects his view that AI will speed up medical progress. However, some in the medical field believe that this forecast is at least a bit exaggerated.
In a recent interview on the with Derek Thompson, CEO David Ricks stated that AI is far from being able to cure the disease.
“If you simply ask them to solve biology or chemistry questions, they aren’t particularly good at it,” he said. “They are trained on human language, not on the language of chemistry, physics, and biology.”
One reason why AI investment has reached record – high levels, comparable to the GDP of some , is the belief that the technology could bring about revolutionary scientific breakthroughs. During President Donald Trump’s press briefing announcing the Stargate Project last year—a $500 billion investment in AI infrastructure through 2029— executive chairman Larry Ellison said that the project could lead to a cancer vaccine, one that could be developed within just 48 hours.
The current reality of AI cancer research
Although Ricks has some doubts about AI’s scientific research capabilities, several AI models have made significant progress in cancer research. For example, Harvard’s Sybil AI model in 2023 lung cancer risk within six years.
And Google DeepMind’s AlphaProteo model has in designing protein binders that target certain molecules, including those related to cancer. In fact, Eli Lilly uses AlphaFold, another AI system developed by Google DeepMind, and maintains a partnership with it.
But Ricks said that current AI capabilities are just a tiny fraction compared with the need for more scientific research. “We can get a machine to predict things quite well, like predicting the structure of a protein,” he said. “But that is perhaps one 1,000th of the problems we encounter in drug discovery.”
The Eli Lilly CEO is betting on tailored AI models to achieve scientific advancements. During the interview, he noted that most LLMs fail to master the subtleties required to handle biology, something he thinks models trained on advanced and specific data could achieve one day.
“The future actually lies in building more and more models for those narrow prediction problems because, unlike human language, biology doesn’t follow all the same rules in the same way,” he said, similar to Google DeepMind’s AlphaFold and AlphaProteo.
Still, Ricks thinks that humans, whether with or without AI, are still a long way from making significant progress in biological R & D, despite the medical advancements that have already been made. “We’re kind of like a toddler in the language of biology,” he said.