AI Superpowers: China, Silicon Valley, and the New World Order

by Kai-Fu Lee

data only becomes truly useful to algorithms once it has been labeled. In this case, “labeled” doesn’t mean you have to actively rate the content or tag it with a keyword. Labels simply come from linking a piece of data with a specific outcome: bought versus didn’t buy, clicked versus didn’t click, watched until the end versus switched videos. Those labels—our purchases, likes, views, or lingering moments on a web page—are then used to train algorithms to recommend more content that we’re likely to consume. Average people experience this as the internet “getting better”—that is, at giving us what we want—and becoming more addictive as it goes. But it’s also proof of the power of AI to learn about us through data and then optimize for what we desire. That optimization has been translated into massive increases in profits for established internet companies that make money off our clicks: the Googles, Baidus, Alibabas, and YouTubes of the world. Using internet AI, Alibaba can recommend products you’re more likely to buy, Google can target you with ads you’re more likely to click on, and YouTube can suggest videos that you’re more likely to watch.