I tend to feel unconfident in my opinions. Partly, it's because I might look bad if my views are wrong. Another reason is that most of the time, the people in the room share similar views to mine. I feel I have little to add. What makes me
AI
Lilian Weng's recent article, "Why We Think", provides an excellent overview of the current state-of-the-art in AI reasoning. Intriguingly, and perhaps unsurprisingly, the methods employed to enable AI to "think" closely mirror human cognitive processes. These parallels are summarized in the table below. Table
AI
While forecasting how AI will impact my career in Consumer Lending Credit Risk Management (Article Link), I explored the timeline forecast methodology used in AI 2027. Prerequisites Time Horizon The concept of time horizon, proposed by the Model Evaluation & Threat Research (METR) team, states that the "50%-task-completion
AI
Recently, I've been exploring the theory of computation. With the rapid advancement of artificial intelligence—essentially a vast collection of algorithms and computational instructions designed to process inputs and generate outputs—I find myself increasingly curious about the fundamental capabilities and limitations of computation itself. Concepts such as
Background Recently, I have been exploring AI 2027 a detailed month-by-month forecast of artificial intelligence advancements through 2027. This resource offers a comprehensive framework outlining how AI is expected to evolve over the next three years. Inspired by this approach, I aim to conduct a similar thought experiment focused specifically
In the offline cellphone-loan business, we have observed that default risk increases with the premium features of the device used during loan application. For example, borrowers applying with phones that have larger storage capacities tend to show higher default rates—counter to the intuitive expectation that “better” devices signal lower
Have you ever clearly explained your viewpoint, watched the other person nod in apparent agreement, and then discovered—hours or days later—that nothing changed? I've experienced this many times, and it taught me an important lesson: communication isn't a single-step process. For words to drive
The "Liberation Day" tariffs represent a significant shift in global economic policy, marking a pivotal moment in international trade relations. To fully grasp the underlying causes and far-reaching impacts of these tariffs, it is essential to employ robust mental models that clarify complex economic, political, and geopolitical dynamics.
AI
Building neural networks often feels like magic. We import libraries like PyTorch or TensorFlow, stack some layers, call .fit(), and voilà! But what's happening under the hood? To truly understand concepts like backpropagation, I decided to follow Andrej Karpathy's fantastic 'nn-zero-to-hero' series and build
搞金融的人应该对利息的概念并不陌生，各类贴现的计算方式可以变得异常复杂。行业内大部分人关注如何运用利息的概念去创造财富，却很少从行为解释的角度深入探讨。最近两年，我阅读费雪的利息理论颇有收获，于是结合自己的理解，整理出了一份文章。 张五常的《经济解释》是我多年前大学时期购买的书籍，距今已有七八年。当时只能勉强理解第一卷中关于科学方法、分工和需求的内容，这部分与亚当·斯密的《国富论》相似。然而，当读到第二卷涉及利息、宏观经济学与财富等较为复杂的内容时，由于难以理解，便中途放弃了。 最近回香港老家，偶然翻阅起这套书，重新激发了我对利息理论的兴趣。这一次重读利息理论部分，我不仅真正理解了内容，也进一步加深了对费雪理论的认识。 利息、风险与交易费用 作者指出，现实生活中存在五花八门的利率，如存款利率、贷款利率、长短期利率和高利贷利率等。这些利率的差异主要源于两个因素： * 通货膨胀预期：市场对未来货币价值的预期影响利率水平。例如，通胀预期升高会导致利率上涨；长期利率一般高于短期利率，这是因为长期未来货币贬值风险较大。扣除通胀预期后得到的「实质利率」(real rate of in
I've recently been reflecting deeply on an important question: what exactly am I busying myself for? To understand this better, I started using Toggl, a time-tracking tool, to log my daily activities. This exercise quickly revealed that a significant portion of my perceived busyness comes from frequent distractions.
Recently, I've started noticing the benefits of quickly picking things up and putting them down again. It's like having a fast startup and shut-down time for my mind, making it easier to switch smoothly between tasks or ideas. Usually, my mind jumps around a lot, juggling