Summary of AI's Side Effects on a Veteran Developer

I suddenly realized it has been a long time since I sat down and wrote a proper technical blog post on Juejin. The technologies I write about now all come from AI, which always makes me feel like they aren't really my own skills. Maybe in the AI era, obsessively grinding at technology really has reached a dead end.

Below are some of my direct feelings from using AI for more than a year. I call them AI side effects.

AI made me lose the ability to think deeply

This sounds a bit counterintuitive: AI learned to think deeply, but I didn't.

In the past, when I hit a stubborn component failure in production, my process was fixed: dig through logs, trace the source code, draw the call chain. I would sit there for half a day, and finally pull out the bug hiding in some corner. That sense of achievement was genuinely satisfying.

Now? I hand the problem description to AI, lean back, scroll through my phone for a while, and wait for the answer to arrive.

The answer comes fast, but I've noticed my “troubleshooting muscle” is atrophying. That intuition for following clues was built by repeatedly digging through source code with my own hands. AI has cut off that path entirely.

Even more obvious is my attitude toward technology. A batch of new terms seems to appear every day: Harness, Agent, Skill. I often glance at a concept, run a demo, and call it “understanding” — just scratching the surface. Over time, I suspect I'm turning into a half-baked programmer: knowing a little about everything and going deep into nothing.

There's another interesting observation. On the company's internal tech forum, posts that purely share one piece of knowledge are getting rarer; everyone is writing in the AI direction. No other reason: if you carefully introduce an ordinary piece of knowledge, what you write may not even be as good as what AI produces. But write about the latest AI developments, and even shallow content gets plenty of traffic. This is probably the most honest signal the market gives.

AI plunged me into self-doubt

I entered the industry around 2018, when technology still came first. As a newcomer, I would grind away at JVM internals, buy piles of books, and spend weekends online watching tutorials. In the era without AI, that kind of learning slowly accumulated into my technical assets. Everything you learned grew on your own body.

But once AI arrived, that logic no longer held.

Learning something new used to give me the feeling of being better than others. Now that feeling is hard to maintain. With AI around, it's hard to say I'm much stronger than a programmer who just entered the industry. No matter how broad your knowledge is, it can't be broader than a model that is always online and has read the entire internet.

My mindset for writing blog posts has changed too. The early pleasure of “finally understanding a problem and being unable to wait to share it” is hard to recover now. It feels like playing a game in god-tier equipment, slicing through everything, knocking down the boss in a few moves, and finishing quickly. It's satisfying, but all that's left is emptiness. You didn't win the game; you won with the equipment.

Everywhere online now promotes “anyone can write code.” To be honest, when programmers see this, isn't there a trace of resentment? After all, you've accumulated skills for so many years, and then AI arrives and stomps that threshold flat.

AI really has made the programmer profession precarious

I think most people won't deny this. Although we're still like the person riding the horse and can still do some commanding, as Agents become more complete and foundation model capabilities keep growing, this horse will most likely run in the direction of “no groom needed.” Whether the reins are still in your hand by then is another question.

The AI era looks full of opportunities, but ordinary people actually have few

There's a popular claim now: an ordinary person only needs a great idea, uses AI to make an amazing work, and can achieve financial freedom and make a fortune.

I think that idea is pretty naive. In the AI era, even if you do create something outstanding, the thing itself has no technical barrier. Anyone can use AI directly to make the same thing as you. Your moat isn't the work; it's that other people haven't thought of it yet. But “haven't thought of it” has a pathetically short shelf life in an era when information spreads so fast.

I often think that our use of AI is like the gold rush in the Americas back then. In the end, the ones who made a fortune were never the gold diggers; they were the shovel sellers.

---

After writing all this, I'm not trying to be doom and gloom. AI is genuinely useful, and I use it every day. It's just that as someone who has been in the industry for nearly ten years, I think some changes are worth recording. This isn't a conclusion, just an ordinary programmer's honest words. If you feel the same way, then I'm not alone.