SOCIAL MEDIA t only twenty-four, Jimmy Donaldson, aka MrBeast, has A created a collection of YouTube channels that collectively boast 221 million subscribers and more than 35 billion total views. Clearly he knows a thing or two about how to scale. One of the things that most stands out to me about MrBeast is how good he and his team are at creating titles for his videos that tell you exactly what you’re going to see in a way that makes you want to watch it. “Would You Sit in Snakes for $10,000?” “1000 Blind People See for the First Time.” “Can 50,000 Magnets Catch a Cannon Ball?” There’s no question what any of those videos are about, and no question that I want to watch them. I mean, aren’t you wonder- ing right now if 50,000 magnets can catch a cannonball? The content is the marketing. Perhaps not surprisingly, then, one of the first things I did when I started thinking about GPT-4’s potential impact on social media was to spend an hour or so feeding GPT-4 various prompts to generate titles for potential MrBeast videos. While GPT-4 produced fairly generic responses much of the time, its 97