Putting Writing First: Navigating AI Anxiety and Returning to Real Thinking

Losing Agency

I love AI as a toy. Late last year, I started using Cursor to build little tools. Most of them ended up in the trash, and only a handful ever made it to the demo stage.

Writing bugs and debugging them gets you hooked fast. Listing ten polish items in a document, feeding them to Cursor one by one, and checking them off as they get resolved—that instant feedback loop is an absolute rush. Even with AI doing the heavy lifting, these were still just dead-simple little utilities. But for someone whose previous coding experience peaked at printing a multiplication table, it felt amazing.

开发项目的页面

Lately, though, I’ve started feeling a real sense of anxiety and emptiness.

On one hand, because I don’t actually know how to code, AI is basically pure magic to me. I type a prompt, it hands me a card; if I don’t like it, I swap it for another. But in the middle of all that, it frequently gets trapped in infinite loops or breaks code that was already working fine, which is excruciating.

On the other hand, whenever I look at my backlog of old ideas through the lens of “productization,” almost none of them hold up. They’re either completely self-indulgent or too vague to pin down. Finding an idea that’s genuinely valuable—or even just interesting enough—has turned out to be far harder than I imagined.

AI represents a massive leap in productivity: an ordinary person waving a magic wand suddenly has skills that used to take years of grinding to acquire. But now you’re standing there holding a massive hammer with no nails in sight. And the harder you scramble to find a nail, the heavier the anxiety gets.

Some people choose to focus on making the hammer bigger. But I don’t think that’s who we are—we’re just using it.

During that confused stretch, I was frantically hunting for nails. I combed through my old notes, scrolled Xiaohongshu (a Chinese lifestyle platform), and even tried sitting in quiet meditation, trying to figure out what might actually make a viable product. None of it sparked that undeniable urge to go build. It only made me more anxious.

Then I remembered an earlier experience that gave me a clue. When I was fully absorbed in doing one thing—building my blog from scratch—a real need surfaced completely on its own. The full blog project had a bloated /node directory. Opening the repo directly in Obsidian forced it to spend ages indexing so many files that every launch took tens of seconds.

Once I ran into that, I found a solution and wrote a plugin in about two hours to fix it: OB File Ignore:隐藏文件并不再索引.

The need arrived naturally, and solving it felt great. I was just following my instincts, letting things flow.

My Approach: Putting Writing First

For 2025, I set two major goals for myself:

  1. Coding
  2. Writing

I’ve decided to flip their priority: put more energy into writing, and let programming ideas surface on their own.

This comes partly from wanting to avoid the anxiety of hunting for nails, but also from what I hope to get out of writing:

  1. AI FOMO easily traps us into obsessing over the tooling. We test various RAG knowledge-base setups and compare Deep Research workflows, forgetting that the real purpose of these tools is to help us learn, research, and think.
  2. Writing is both the process of thinking and the result of it. What AI gives me, on top of basic coding ability, is dialogue, deep thinking, dissecting arguments, and structuring logic. Instead of pair programming (not that what I do really counts as pairing yet), I should be exploring what “pair learning” and “pair writing” can look like.
  3. I’m someone with wide-ranging interests but terrible follow-through. Digging into any topic takes a lot of factual input and finding how it connects to me personally. Too often, the initial excitement fizzles out right at the start, leaving me only ever scratching the surface.

This could turn out to be the best of times.

Comments