Deprived of thinking?
This post is a rant.
I’ve been having this mindset in the past few days, reflecting to myself that my thinking patterns have drastically downgraded due to being more and more used to AI-assisted coding.
Although I am not an “ask ChatGPT/Claude/Grok for everything” kind of guy, I have increasingly found myself thinking slower on coding questions, and having this expectation of an AI to auto-complete for you every time; this is especially prominent when I do online assessments (due to me getting graduated and are looking for jobs). To be clear, I am a 4+ year Olympiad in Informatics veteran; LeetCode medium/hard questions are generally within the “I have seen this problem before” scope for me. Still, I increasingly find myself not having the speed to come up with the optimal solution in a short time.
Another thing I noticed is that my writing has become more and more AI-ish, i.e. adapting those languages favoured by LLMs that are actually bad. Not as obvious as “not X but Y,” though. I am not a linguistics expert, so I can’t generalize what “AI-ish” is about, or why I think my writing has become AI-ish; it’s just a vague feeling that what I’ve written reads dull and mechanic. At first, I thought it’s just that my own writing before has a melancholy and literaturistic/artistic taste, which has become washed away since I communicated with AI too much. But now I think those can also be contributed to me thinking in a more and more superficial way—I have nothing thoughtful to express, that’s why.
This rant is real. Deep and patient thinking is what defines me, my very proud feature that largely differentiates with others. For every questions I encountered, I would think about it before coming up with an answer, and in the process, it keeps my brain cells active. Now that I’m used to commanding AI agents, I feel like I am a product manager, commanding employees to solve puzzles in my stead. My brain cells are dying, and at a worrying speed.
What made me rage-realize my degrading of brain cells, is a frustrating experience last week. I participated an online assessment from a San Jose tech company (unnamed but you know which company it is). Four LeetCode easy/medium questions. While I solved the first three questions rather effortlessly, I was getting stuck in a wrong mindset when solving the fourth question which involves a bit of greedy paradigm. Well, who can I blame for not getting the optimal solution in one hour but myself? It’s not a LeetCode hard. Greedy algorithms are the most diverse and tricky—practice doesn’t make you perfect on this (unless you actually solved the exact same question before), smartness does. It means that I am no longer that smart, and I have to admit it. What if my IQ has fallen below 120? That would be a disaster if true.
Okay, I just did a quick test at mensa.org and got . Now I am relieved. I know it’s not very accurate as it’s just 35 questions rather than a full Mensa test suite, but I just want to confirm I am not that dumb right now.
Still, from this same test I’ve found similar thinking patterns when I encounter questions I can’t think of an answer, at the end of the test when the end of timer is nigh. Instead of thinking hard to find out, my anxiety is off the charts and it really makes me hard to focus on problem-solving. Guess I really have to work on that.
Now that my rant is over, here are some conclusions I should work on in the future.
- I should do more intellectual problem-solving as a daily habit, and preferably with a timer. This is currently my weakness and I must use this task to retrain my brain cells (and overcome my anxiety when running out of time).
- For my mental health, I really should write and reflect more of myself, in my own words and without the help of AI. It’s really inefficient not to auto-complete with AI agents when you write code; but for the problem-solving scenarios, maybe I should set some boundaries—e.g., solve the problem unassisted if I find it hard or thought-provoking, then check the answers with AI agents and only let them implement.
- I should really look for a PhD and dwell more on improving my expertise. The AI is gradually taking over people’s jobs in the following years, and no one can stop this trend—at least what the activists do cannot. Instead of living in this constant fear of getting replaced by AI, I should just arm myself with expertise that are hard to be replaced by AI in the near future. Although people still say that getting a job in the Bay Area is the correct path, I don’t find it safe anymore. Earning a living as a Software Engineer is as easily replaceable as a 2D artist, or an Uber driver—lose of jobs is happening in real time as we already see.
Looking back at this blog post, I see my anxiety off the charts in the text I just wrote. That may not be a bad thing actually—getting anxious about my own career is probably better than not thinking about it at all. I just have to vent it off sometimes, like this post.