I’ve spent 30 years trying to harness AI to fight cancer. I’m dismayed by the AI pseudo-revolution

On a hot day in the summer of 1993, my academic supervisor handed a the seminal paper by David Rumelhart and Geoffrey Hinton and simply said, “Implement this!” At the time, we used programming languages like Modulo-2, Pascal, and occasionally ANSI C. But hardly anyone had heard of “backpropagation” — a learning mechanism for AI that, to this day, remains foundational for training artificial neural networks on complex tasks.

Around that same time, my grandfather passed away from lung cancer. A former prisoner of war, he had introduced me to both Eastern and Western philosophy, despite never having earned an academic degree.

So when I began my Ph.D. in 1996 in post-unification Germany, I knew exactly what I wanted to do: harness AI to fight cancer.