A year ago today, I was eating a Princi cream bun at Milan Cathedral, admiring the Duomo, and casually scrolling through news about a major power outage in Spain while lying in Sempione Park. That night I was supposed to fly back to Barcelona, and I was wondering if Barcelona Airport would even reopen. Now, this year, I’m reviewing questions submitted by students from the probability TA course I’m assisting with, and rushing to finish my machine learning assignment. However, all machine learning assignments come with a lot of fragmented free time. This model I’m working on takes about 10 minutes to train, and there are many different variants to prepare. After looking at the model, I need to check papers; after checking papers, I need to assess data reasonableness. Even if something seems unreasonable, I’ll discuss it with AI to make it sound plausible. Finally, I also have to keep the report within six pages. I quietly deleted some experimental results that seemed unimportant but were interesting to me, feeling a bit empty.

While waiting for models to converge, I’m always distracted. Even when I am focused on the current task, I dare not change too much code at once. I’m afraid that if I make one move, the task running in the terminal will start bugging out, and one wrong step will lead to many, making my deadline catch-up plan even more frantic. So I start scrolling through some meaningless content on my phone, trying to numb myself. By the time I’m done scrolling, the model is usually ready. Then I look at the results, analyze the data, continue, and perhaps read a paper or something. This constant switching, integrating, and thinking often makes me feel like I’m fighting a solitary battle.

But actually, it’s not like that. My schedule today was like this: I slept for over nine hours (I know this is an unreasonable amount of time for a graduate student, but I was just very tired and love sleeping). I went to record a podcast and had lunch with my good friend, attended a lab meeting, and also had a delicious dinner. But it’s when I settle down and try to deal with lab work, TA duties, or coursework that I often question myself. Will everything I’m currently doing be shallow and meaningless? Even if I really discover something, in the deluge of AI, I feel like a tiny droplet. Perhaps there’ll be a small splash when I drop, but AI and those who expertly wield it will always continue to produce even more distracting content. Who besides myself will genuinely care about my research and ideas? In an era where attention is increasingly fragmented, too many people judge others solely by their achievements, or even dismiss achievements and judge character based on appearance or rumor. So what’s the point of all this LLM research, teaching work, or even just doing regular homework that I’m currently engaged in?

I also wonder if LLMs were to conduct this research themselves, they might do it very well. In an era where LLM knowledge is gradually surpassing human knowledge, we can casually discuss content that would have taken me a considerable amount of research to understand. So, what is the meaning and value of knowledge then?

Actually, I’ve thought about these questions many times before and have some preliminary answers. For example, I am currently creating “dots,” and these dots will be connected in the future. Like when I successfully combined past internship experiences to secure a better internship, or when my past ideas perfectly merged, making me want to shout “Eureka!” But just now, while brainstorming randomly, it suddenly occurred to me that connecting my accumulated knowledge capital should allow me to gain more freedom and mobility.

My life function is this: my life’s maximum utility function is the sum of happiness from this very moment until the end of my life. And it’s time-weighted, so the utility of happiness now or in the short term is higher and more easily obtainable. But the future is still long, and even when time-weighted and discounted to the present, it remains important. Therefore, I cannot simply abandon academics and research now, because they are crucial for the future.

So why are academics and research, and even my current assignments, important? We need to work backward. If I want to pursue happiness, I believe the best ways are either to accumulate more diverse experiences to enrich life, or to find joy through pleasant interactions with people. It means accumulating my options, and these options are “cost-free to acquire, but potentially very joyful to use at some point.” So, essentially, accumulating options ensures that in the future, I will always be in a state of “having many choices.” This state, if I choose well, can bring me happiness. As for how to increase my choices? I believe the sources can be categorized into two: money and freedom (mobility). These are also what I currently most want to balance. If I sacrifice short-term happiness now, it should be in exchange for long-term sources of happiness, namely money and freedom. And the experiential paths I’m taking by doing things that don’t make me happy right now (like my current assignments) are usually aimed at improving these two indicators.

There are many ways to acquire money and freedom, but increasing the options one can choose from is crucial. For example, a popular recent question to ask AI is, “What completely different job do you think I would excel at?” The AI told me that “Non-fiction Narrative Curator” or “Urban History Interpreter” would be a great fit for me. I found its reasoning very compelling, but how can I comfortably try out these areas with a backup plan in place? Naturally, first, I need to accumulate enough freedom and money to make these choices. So, in principle, building a personal brand, accumulating coffee chats, and creating podcasts can help me meet more people, giving me more options. This is a direct contribution. Then, becoming a software engineer and trying to secure a Google internship are excellent ways to accumulate both money and freedom. And managing a personal media platform also boosts my freedom because if I am capable enough (including increasing money and freedom), I can move to different cities and spend more time doing what I want (things that directly bring happiness), allowing me to, for instance, go back to Milan, lie down, people-watch, and still maintain an income.

Well, after circling around, I still have to return to the deadline I’d rather not face. The LLM skills I’m learning now, or the fundamentals of machine learning, besides bringing me the intrinsic joy of learning, are also to demonstrate my capabilities. Accumulating more such “dots” will lead to more opportunities in the future, and choosing good opportunities can bring more money and freedom. Money, freedom, and opportunities are cyclical and interconnected. And my current investment in skills, in the long run, can at least bring me one of these: freedom, money, or opportunities. The combination and positive feedback loop of these three can also bring me happiness. For example, I can more seamlessly explore the field of “urban history interpreter,” bringing me more happiness and a positive cycle. If the sum of these three is increasing, I should consider myself lucky and on an upward trajectory.