Eli Lilly CEO sees AI as “not particularly good” at problems in biology or chemistry

Now, a growing group of technology leaders is hailing artificial intelligence as the key to solving a medical mystery that has puzzled doctors for thousands of years. That’s what Google President Ruth Porat anticipation Last October. That’s why Anthropic CEO Dario Amodei coined the term “Compact 21st century“, which reflects his view that artificial intelligence will accelerate medical progress. But some in the medical field believe these expectations are at least somewhat exaggerated.

In a recent interview on Plain English Podcast With Derek Thompson, Eli Lilly CEO David Rex said that artificial intelligence is far from curing the disease.

“If you just ask them to solve problems in biology or chemistry, they won’t be particularly good at it,” he said. “They are trained in human language, not in the language of chemistry, physics and biology.”

One reason why investment in AI is at record levels, rivaling the GDP of some Developed countriesIt is the belief that technology can enable revolutionary scientific breakthroughs. During a press conference announcing President Donald Trump’s “Project Stargate” last year, a $500 billion investment in artificial intelligence infrastructure through 2029, oracle CEO Larry Ellison He said The project could lead to a cancer vaccine, one that can be created in just 48 hours.

The current reality of AI cancer research

While Rex has some doubts about the scientific research capabilities of AI, several AI models have made significant advances in cancer research. Harvard’s Siebel AI model in 2023, for example, I predicted accurately Risk of developing lung cancer within six years.

Google DeepMind’s AlphaProteo model did just that Proven effective In designing protein ligands that target specific molecules, including those associated with cancer. In fact, Eli Lilly uses AlphaFold, another AI system developed by Google DeepMind, and maintains a partnership with it.

But Ricks said current AI capabilities are just a drop in the bucket compared to the need for additional scientific research. “We can have a machine that predicts things well, like predicting the structure of a protein,” he said. “But this is probably one-thousandth of the problems we face in drug discovery.”

Eli Lilly CEO is placing his bets on artificial intelligence models designed to seal the deal for scientific progress. During the interview, he noted that most LLM students fail to master the nuances required to engage with biology, something he believes models trained on advanced, specific data could one day achieve.

“The future here is actually to build more and more models for these narrow prediction problems, because biology, unlike human language, doesn’t follow the same rules in the same way,” he said, similar to Google DeepMind’s Alphafold and AlphaProteo.

However, Rex believes that humans, with or without artificial intelligence, are still far from biological evolution, despite the progress already made in medicine. “We are like a baby in the language of biology,” he said.