He’s sweating $7.3 billion worth of teeny-tiny details.
Alexandr Wang, 25
SCALE AI
As a student at the Massachusetts Institute of Technology, Alexandr Wang was the casualty of a wickedness that plagues college campuses the world over: His roommates were stealing his food. Specifically, they were taking his yogurts. “Yogurt is one of the easiest foods to steal, because it’s so perfectly contained,” he says. “It’s the perfect crime.” So Wang put a camera on the inside of his fridge that he hoped would not just record the happenings of the fridge for him to review later, but know—on its own—that a yogurt had gone missing. He soon realized, though, that such a system would need to learn quite a lot to do what, for a human sentry, would be a relatively simple task. For one, the system would have to learn what a yogurt looks like—and, just as important, what it doesn’t look like. After all, plenty of other things are also perfectly contained. It would also have to learn the numerous places the yogurt might be in the fridge, and the copious configurations of surrounding items. To gain this artificial intelligence, his system would not only need a wealth of images to learn from; it would need the important parts of those images to be labeled. This idea of labeling is key. Only then could the AI learn the nature of the relevant elements. Only then could the data become useful. Indeed, only then could his system spot yogurt theft.
Wang is guided by a core principle: And pervasive though they may be, fridge infractions would not have fueled him forever. But Wang had been studying artificial intelligence, and even had been accepted to Y Combinator’s prestigious accelerator program with an AI company in mind. His fridge experiment made him realize just how essential—and underserviced—labeling was to transforming companies’ wealth of data into beneficial uses. Labeling could empower projects that could help fix far bigger problems, like climate change and expanding medical care access—that is, solutions worth giving a shit about. After his freshman year, he dropped out of college, and at
