AI is a process to convert solar energy to intelligence

 If you’ve ever tried to navigate New York City with an outdated subway map, you’ll know how quickly a “helpful guide” can turn into a recipe for getting lost. That’s because a map is not the territory . It’s just a compressed, simplified way to describe reality. And here’s the kicker: there isn’t just one “correct” map—there are multiple. A street map, a restaurant guide, a real-estate zoning plan—each describes the same city but with a completely different lens. Sometimes, the map even changes the territory itself. Just think of NYC’s city planning: decisions drawn in ink eventually get poured in concrete. The same is true for how we understand the world. Our mental models are just approximations of reality. None of them are perfect. Each comes with blind spots, biases, and assumptions. Why do we cling so hard to different perspectives? Often it’s because of incentives, baked-in human biases, and the gravitational pull of group identity . "Us vs. them" isn’t just a tribal...

 If you’ve ever watched a child stubbornly choose candy over carrots, you’ve witnessed evolution whispering through human behavior. Children, by nature, optimize for short-term convenience and pleasure rather than long-term gains. It sounds irrational to us “grown-ups,” but if you rewind to the hunter-gatherer era—where life expectancy was short and tomorrow was never guaranteed—grabbing joy in the moment was the smart move. Live fast, have fun, survive the day. Evolution coded that into us. Fast forward to the last hundred years: technology has rewritten the rules of life. We now routinely live decades longer than our ancestors, but evolution’s pace is glacial. Genetic change takes countless generations, while medicine, AI, and silicon chips advance in decades. Put differently: our bodies and instincts are still running “Stone Age 1.0,” while our technology is already in “Silicon Age 10.0.” That lag explains why we often feel out of sync with the modern world. This mismatch is a...

 When it comes to making decisions under uncertainty, the biggest trap isn’t the lack of information—it’s the illusion of precision. Computer models may promise exactness, but more often than not, they offer false comfort and open the door to catastrophic mistakes, as history has shown in both insurance and investing. The wiser path is to resist two natural tendencies: discounting simply because of ambiguity, and assuming others know more just because you don’t. Instead, think critically about what’s truly knowable, and when odds look unusually favorable, don’t shy away from taking a calculated gamble. In short, uncertainty isn’t your enemy—misplaced confidence is.

 In psychology, “peer group influence” and “self-categorization” explain why people sharpen the contrast between “us” and “them”—we not only adopt the norms of our own group but also exaggerate the differences with outsiders. Strikingly, this mirrors what happens in computer intelligence. Just as digital image processing uses algorithms to enhance clusters and make a blurry photo more distinct, the human mind amplifies social boundaries to bring order to a messy social world. Similarly, data mining highlights subtle correlations by separating meaningful patterns from noise, much like teenagers fine-tuning their identities to align with friends while distinguishing themselves from rival groups. In both cases—whether in silicon or in neurons—the goal is clarity: turning ambiguity into structure, and noise into meaning.

 Modern programming, for all its brilliance, has become tangled in its own complexity. Too often, developers spend more time wrestling with obscure libraries, fiddling with environment setups, or deciphering someone else’s coding quirks than actually building ideas. Enter Generative AI: a liberator from these nuisances, allowing us to bypass the grunt work and concentrate on what really matters—the logic, the creativity, the spark of innovation. After all, computer languages have always been about communication first. From punch cards to Assembly to C and beyond, interpreters have been steadily nudging code closer to natural language. Now, with large language models, we stand at the edge of a fascinating shift: for the first time, our everyday words themselves might become a true computer language.