Algorithms! We hear a lot about them. They drive social media platforms and, according to popular understanding, are responsible for a great deal of what’s wrong about the internet today—and maybe the downfall of democracy itself. But … what exactly are algorithms? And, given they’re not going away, what should they be designed to do?
Evelyn Douek and Quinta Jurecic spoke with Jonathan Stray, a senior scientist at the Berkeley Center for Human-Compatible AI and someone who has thought a lot about what we mean when we say the word “algorithm”—and also when we discuss things like “engagement” and “amplification.” He helped them pin down a more precise understanding of what those terms mean and why that precision is so important in crafting good technology policy. They also talked about what role social media algorithms do and don’t play in stoking political polarization, and how they might be designed to decrease polarization instead.