Bernhard Schölkopf

There are researchers who solve one problem and move on to the next. And then there is Bernhard Schölkopf, who solves a problem, turns it into an entire field, and then asks whether that field is looking at reality the right way. Born in Germany in 1968, he studied physics and mathematics before landing in computer science, a combination that gave him an unusual perspective: that of someone who not only wants algorithms to work, but wants to understand why they work.

His most celebrated contribution came when he took support vector machines (SVMs), which Vladimir Vapnik had developed for linear data classification, and extended them into the nonlinear world using mathematical functions called "kernels." That extension allowed SVMs to learn patterns in biological, medical, and climate data of a complexity that had previously been out of reach. He then led the creation of an entire field of kernel methods, unifying under one mathematical framework dozens of algorithms that had previously lived in isolation. In 2018 he received the Leibniz Prize, Germany's highest scientific honor, for that work.

But perhaps his most ambitious contribution is the one he is building now, combining machine learning with causal inference. Most AI models detect correlations: they know that A and B appear together, but they do not know whether A causes B or the other way around. Schölkopf decided to attack that problem head-on, developing methods for machines to recognize causal structures from observational data. This work has direct implications in medicine, economics, and any field where a wrong decision can cost lives. In 2022 he received the ACM-AAAI Allen Newell Award, one of the most prestigious recognitions in artificial intelligence, and he is a Fellow of the Royal Society.

What I find fascinating about Schölkopf is that he never settled for AI that simply "works." He wants AI that understands. And that distinction, between a system that predicts and one that reasons, is exactly the conversation everyone should be having today. Every time you use an AI tool that not only gives you an answer but can walk you through the reasoning behind it, a little of that obsession of his is pushing from underneath.