Why Aren’t AI Drugs Available Yet?
▼ Summary
– Peter Ray’s childhood trauma of losing his mother to cancer inspired his career as a medicinal chemist dedicated to developing cancer drugs.
– Drug discovery is a slow, painstaking process involving targeting proteins and designing molecules to interact with them.
– Many potential drugs fail during testing due to unforeseen side effects or ineffectiveness, often after years of work.
– Only a small fraction of drug candidates progress through all clinical trial phases, with over 90% failing along the way.
– Success in bringing a drug to market is rare, with researchers like Keith Mikule describing such achievements as being “unicorns.”
The journey from lab to pharmacy shelf for new medications remains painfully slow, despite revolutionary advances in artificial intelligence. While AI promises to accelerate drug discovery, transforming how we develop treatments for diseases like cancer, the reality is more complex. The process still involves countless hours of meticulous work, heartbreaking setbacks, and unpredictable biological hurdles that even the most sophisticated algorithms struggle to overcome.
Peter Ray understands this challenge intimately. Growing up during Zimbabwe’s liberation war, he witnessed his mother’s losing battle with cancer, an experience that shaped his life’s mission. Now a lead drug designer at Recursion Pharmaceuticals, Ray reflects on his childhood promise to make a difference. “That commitment stays with me,” he says. “Getting effective cancer treatments to patients drives everything I do.”
Traditional drug discovery resembles solving a microscopic puzzle with constantly shifting pieces. Scientists first identify a target protein, studying its intricate folds and crevices where potential drug molecules might bind. Using computer models, they design compounds atom by atom, hoping to create the perfect molecular key. But biology rarely cooperates with theoretical designs. When tested on living cells, most experimental drugs fail spectacularly, sometimes harming healthy tissue or simply not working as intended.
Keith Mikule, a biologist at Insilico Medicine, recalls one particularly disheartening project. After five years of work involving dozens of researchers and thousands of molecules, their most promising candidate revealed dangerous side effects. “All that effort, and we had to start over,” he explains. Such stories are common in pharmaceutical research, where failure rates exceed 90% at various development stages.
The clinical trial process introduces further obstacles. Even when a drug shows promise in lab tests, it must pass three rigorous human trial phases, first proving safe in healthy volunteers, then demonstrating effectiveness in small patient groups, and finally confirming benefits across diverse populations. Most candidates falter somewhere along this path. Mikule, part of the rare group with a successful drug launch (niraparib for ovarian cancer), describes successful researchers as “unicorns” in this high-stakes field.
While AI tools help identify potential drug candidates faster than ever before, they can’t yet predict how complex biological systems will respond. Human biology’s sheer variability, from genetic differences to environmental factors, creates challenges that still require human intuition and years of testing. The emotional toll on researchers is immense, as each failed compound represents years of work and, for people like Ray, personal hopes for helping patients like his mother.
What keeps scientists going through this grueling process? For many, it’s the knowledge that every failure brings them closer to breakthroughs that could save lives. As AI continues evolving, researchers cautiously hope it might one day crack biology’s code more reliably. But for now, the path from concept to cure remains a marathon, not a sprint, filled with equal parts frustration and determination.
(Source: Wired)