OpenAI's 12-day livestream, with its continuous surprises and the official launch of Sora today, has pushed the long-dormant AI buzz to a new high. Having used AI for nearly a year, I've felt the many conveniences it brings to my work and life. Today, I'd like to share some of my recent tips and insights on using AI. Two recent feelings about AI First, your mastery of new tools determines your future class. What is AI? Few people deeply ponder this. If we liken AI to a stick, you'll find that whether you're a beggar, a porter, Archimedes, or the Monkey King, the stick itself doesn't change, but different users can unleash different powers. Also, even for the same porter, a more convenient and better stick can increase efficiency and enable more work. So, in what dimensions can AI help you improve? I think there are roughly four: Data dimension: Efficiently collect, clean, integrate, and analyze massive data, giving individuals more comprehensive, accurate, and real-time information support for decision-making and insight. Knowledge dimension: Integrate knowledge systems and cutting-edge information across fields, helping individuals quickly acquire, understand, and internalize new knowledge, thereby expanding cognitive boundaries and deepening thinking. Model dimension: Automatically build, train, and optimize various analytical models, enabling individuals to understand complex problems and formulate efficient strategies with lower trial-and-error costs and higher precision. Experiment dimension: Rapidly simulate and test multiple solutions, allowing individuals to validate hypotheses and iterate with limited resources, shortening innovation cycles and improving outcomes. Ultimately, AI lets us redefine ourselves and find new ways to coexist with technology. The wisdom of using the stick is more important than the stick itself. When you master more advanced productivity tools, you'll naturally show different strengths compared to others. The second feeling is that AI is once again leveling class disparities. Human progress is essentially a process of equalizing scarce resources: the Industrial Revolution, with division of labor and mass production, began to equalize goods: you, the US president, and African children all drink the same Coca-Cola, and crucially, at similar prices. The Internet revolution began to equalize information: you and I can see the same news and information simultaneously, eliminating asymmetry and information gaps. When the AI revolution arrives, it begins to equalize knowledge: you and I can both master professional knowledge in any field. What does that mean? As long as you ask, AI will provide all the knowledge you want. In an era of knowledge equalization, obsessing over degrees is meaningless. AI no longer makes knowledge a barrier to class; whether you're admitted to Tsinghua or Peking University or you're a middle school graduate, the knowledge we master is the same. In the AI era, asking questions is more important than solving them, so you should spend more time not on learning but on training and exploring:

  • Training your thinking ability, problem-solving skills, and ability to communicate with this era's groups when you have all information.
  • Exploring all possibilities of what you pursue and love. In the AI era, when material, information, and knowledge supplies are no longer scarce, people should look inward, exploring their inner drives: your passions, why you live, and what you pursue. As individuals, the challenge in the AI era is: How to acquire the ability to master AI? I think you should start using it, whatever AI it is, and apply it to your work. You'll see a significant productivity difference compared to not using AI. For example: the other day, when chatting with Mr. Zhong of Chongqing Jiecang Wangou, we discussed Chongqing's consumption potential. I used Perplexity to quickly search for Chongqing's population, consumption, Engel coefficient, and urban-rural differences. It summarized all the data for me almost instantly. Then I asked about category management, and it immediately provided all core information on retail category management, making me an instant expert. Without AI, as a professional, searching, organizing, and presenting effective results—even if I could find them among the vast articles on Baidu—would take at least two hours or more. But now I don't even need two minutes. That's the knowledge and information gap AI creates. It lets you master the most needed knowledge in the shortest time; you don't need to pre-learn or memorize. The only thing you need to do is ask a high-quality question in that context. So, what I most want to tell you is, no matter which AI you use or in what scenario, make sure to use it first. Only by using it will you gradually find application scenarios in your work. The next question is: How to ask a high-quality question? I posed this to GPT-o1 Preview, and after 9 seconds of thinking, it gave me the following answer. I think GPT wrote it better than I could, so I won't elaborate here, but I believe points 2 and 3 are particularly important. If possible, definitely use GPT I've tried all AIs, and my current combination is: Perplexity + ChatGPT + Get Notes + Kimi
  1. For deep answers, I personally find ChatGPT the best. A few key points for using GPT:
  • Definitely subscribe to the membership, because o1-preview's chain of thought is very powerful; it helps you think systematically, covering points you might miss, with comprehensive thinking, thorough reasoning, and rigorous logic.
  • Make good use of custom instructions in the custom settings. Custom instructions refer to what you want it to know about you, what kind of person you are, and how you want it to respond. This is crucial for the quality of your dialogue.
  • GPT has memory; you should let GPT remember information, knowledge, and questions you'll need in the future. After a valuable conversation, tell it, "Remember what was said earlier; I might need it later." It will remember your focus areas.
  • Try to communicate on one topic in a single chat window, so it can understand your needs deeply and comprehensively with context.
  1. Perplexity is excellent for search; its information integration is powerful, especially for economic data, book viewpoints, and multi-dimensional information. It's accurate and user-friendly, basically ready to use. For searches requiring internet, you can use Perplexity, or domestic options like Nami; the UI and logic are almost identical to Perplexity, but for accuracy and rigor, I personally recommend Metaso Search among domestic ones.
  2. For fleeting ideas that need quick recording, I usually use Get Notes. It helps me quickly logicalize and systematize unorganized information, then automatically syncs to Evernote.
  3. Kimi I use for quickly reading articles. For less important articles where I just need to browse information, I forward them to the Kimi mini-program, and it quickly summarizes and extracts key points, saving reading time.
  4. This is my current AI tool combination. Of course, I have other AIs on my phone, like Poe, Merlin, Doubao, Yuanbao, Tongyi Qianwen, etc., depending on personal preference. What do I use them for? 1. Extending thinking triggered by related information in reading. When reading, I discuss topics and conceptual extensions with GPT, and use Perplexity to expand knowledge. 2. Basic knowledge in unfamiliar industries. For example, healthcare, fitness, sports, astrophysics, philosophy—anything you're interested in but not your specialty—ask anytime. 3. Reading, summarizing, and organizing content. For instance, for Berkshire Hathaway's 20-year financial reports, I need to extract information. I feed the PDF to AI and ask it to summarize McLean Company's financial data and development patterns from those reports. 4. Helping refine my thinking on certain issues. For example, I've been thinking about human needs and behavior models. I repeatedly discuss with GPT what points haven't been accurately described or exhausted. 5. Deep research in professional fields. For specialized research on global, local, macro, micro, industry, and humanities topics, the AI combination is a powerful assistant. Thoughts on AI's impact on humanity's future At the New Distribution August conference this year, during the AI warm-up session, I shared some views on AI's impact on humanity's future. Surprisingly, they were very similar to many points in futurist Ray Kurzweil's new book "The Singularity Is Nearer: When We Merge with AI," published on June 25, 2024. AI can help humanity in three stages:

First stage—Copywriting support: providing information Second stage—Strategy support: assisting decision-making Third stage—Action support: helping solve work

  1. Sufficient supply of energy, computing power, and data
  2. Distributed, decentralized model applications
  3. Coordination of productivity and production relations The US indeed has a huge advantage here: AI was born there, and labor shortages and manufacturing hollowing-out give the US strong incentives to replace human labor with machines. Just as the Industrial Revolution was born in Britain, essentially because of labor shortages, capitalists were willing to invest in machines to replace humans. More importantly, in the future, AI will penetrate not only manufacturing but also the primary and tertiary industries, with massive replacement. So, from a macro perspective, I'm worried about China's future. We shouldn't bet the nation's fortune on low-margin manufacturing but should make overwhelming investments in technology and information technology. This is counterintuitive. I'm pessimistic about the fate of most humans on this planet, including myself. Fortunately, we are pre-historic humans (humans before AGI), and humanity is not prepared to face such immense power for the first time. Our education, culture, and even civilization are essentially results of humans adapting and compromising to accommodate massive productivity increases. At the philosophical level, humanity hasn't provided corresponding spiritual guidance, meaning we're putting a bullock cart harness on a rocket engine without knowing where the rocket will take us...