“A wealth of information creates a poverty of attention.” Herbert Simon, Nobel Prize-winning economist When Google arrived, it did something quietly radical. A billionaire and a schoolkid typing the same words got exactly the same results. Information became the great leveller, and the only real advantage was knowing how to search. Google took that skill seriously enough that its search scientist, Dan Russell, ran a free course on it, and more than 4 million people took it. AI looks like the same deal. Everyone gets the same friendly chat box. Behind it, though, there are very different brains. On the free and standard plans, the tool quietly decides how hard to think about your question, and most of the time it picks the quick, cheap answer. Pay up to $200 a month and you get a model that’s allowed to slow down, research, check its own work and sometimes find a better question than the one you asked. Two people can type the same words and get different depths of thought depending on their subscription. For the first time, thinking itself is metered. Sociologists have a name for what tends to happen next. In 1968 Robert Merton described the Matthew effect, named after the biblical line about the rich getting richer. He noticed that famous scientists got more credit than unknowns for exactly the same work, so the famous got more famous. A thinking divide compounds the same way: better thinking gives you better work, better work brings in more money, and more money buys better thinking. I talk a lot about compound curiosity. The uncomfortable flip side is that compound thinking can now be bought. But money may not be the biggest divide. In 2022, Google paid Apple $20 billion to stay the default search engine in Safari, and Microsoft had offered Apple 90% of Bing’s advertising revenue to take its place. Anyone can change their search engine in a few taps, yet two of the richest companies on earth were betting billions that almost nobody would bother. Behavioural economists call this the default effect, and AI is no exception. Plenty of people paying for clever models never switch off the automatic setting, and get System 1 answers from a System 2 machine. So there are two ways across the divide. The first is money, and if you’re paying, actually use what you’ve bought: switch off the automatic setting when a question matters. The second is free, and it’s the skill Dan Russell spent years teaching people for Google. Ask better, more thought-out questions; tell the AI what’s at stake, what you’ve already tried and why the obvious answer won’t do. Even a free model gives you far more when you’ve done some of its thinking for it. A billionaire and a schoolkid can still type the same question into AI. The billionaire’s model will think harder about it. The schoolkid’s only way to keep up is to think harder before typing. If you want to ask better questions and be able to look at problems and solutions from another angle, our Behavioural Economics course is a great place to start. Plus, it’s taught by Rory Sutherland, so it’s hilarious as well as helpful!