Sunny

← All weeks

What I’ve been reading (W15, 2025)

Power of people helping LLMs, understanding tariffs, asking good questions, and learning to move fast

  1. LLMs like ChatGPT have flipped the usual story. Usually, big tech breakthroughs start with governments or big companies and slowly reach regular people. This time, it’s the other way around. Every day, people are the ones getting the biggest boost. With just a phone or laptop, anyone can now write, code, translate, learn, or create things that used to take a whole team of experts. That’s not a small step; it’s a massive jump in what one person can do independently. Big companies are still figuring out how to use it, but regular folks are already running with it. For once, the most potent new tech landed in the hands of the people first. https://x.com/karpathy/status/1909308143156240538
  2. A good answer reduces uncertainty and respects our limited attention. But AI now floods us with cheap, convincing answers, making it easy to stop asking. It’s like when the camera was invented in the 1800s. Suddenly, painters no longer needed to copy reality; the camera did it better and faster. That forced artists to shift the frame from realism to interpretation. Impressionists like Monet asked new questions - what does a moment feel like, not just what it looks like? AI is today’s camera. It doesn’t kill thinking—it forces us to evolve it. The value now lies in powerful questions that change how we see, not just what we know. https://platforms.substack.com/p/when-answers-get-cheap-good-questions
  3. Tariffs aren’t just about protecting American jobs. They’re being used as a strategy to isolate and weaken China. China depends heavily on exports to keep its economy stable, while the United States is the world’s biggest consumer and doesn’t rely on imports for survival. If the U.S. puts tariffs on all countries, it forces them to choose between trading with China or the U.S. Most will choose the U.S. because that’s where the demand is. This cuts China off from global trade, hurts its economy, and could eventually shake the political system that’s built on delivering steady growth. Its economic pressure was designed to break China’s global influence. Countries like Vietnam, the Philippines, and Indonesia may even be offered a chance to replace China in the global supply chain, but under rules set by the U.S https://x.com/eastdakota/status/1909822463707652192
  4. One of the reasons public LLMs relatively seem to have a lesser impact within organisations (for now) is the same reason Google Enterprise Search failed. Consumer and business use cases are two different problems. Public LLMs and Public Search work on public data. Organisations have complex sets of disparate data sources with complex permissions structures unique to each business. Cracking this is key to succeeding within businesses. These are boring problems to solve. Boring things bring in more dollars. https://x.com/rajuv/status/1909462837107323226
  5. A good reminder of Clock speed with small teams. Most founders move fast. The great ones break time itself. The best ones reshape physics. Where others seek consensus, they ship and course-correct in hours. While teams perfect one approach, they’re running five. What feels like an impossible execution to others is just an unbearable delay to them. Every milestone spawns three new ones. They’re not racing competitors—they’re racing time itself. Some learn to modulate this tempo, pressing for impossible speed without breaking their teams. But the drive never slows. The urge to bend reality faster than it should move never leaves. The rage at a normal pace never fades. https://writing.nikunjk.com/p/clockspeed and https://sriramk.com/clock-speed