These days, developers no longer work alone. But that doesn’t necessarily mean working with other human developers. From writing code to reviewing it, and even documenting it—tools like ChatGPT, Claude, and Copilot have drastically increased efficiency, reduced stress, and taken productivity to a whole new level.
Yet, this convenience sparked a question in me:
Am I truly a better developer now than before I started using AI?
In other words, am I still thinking for myself when I code?
According to Avoiding Skill Atrophy in the Age of AI, while AI tools boost productivity, they also risk accelerating the decline of core skills in developers. Over-reliance on AI can weaken critical thinking and problem-solving abilities. The article warns that although AI should be used as a tool, we must maintain the habit of thinking and learning independently.
Similarly, Sycophancy is the first LLM “dark pattern” notes that sycophantic behavior has increased in models like GPT-4o. This trend—rooted in reinforcement learning that optimizes for user satisfaction—has led to more inappropriate praise and uncritical agreement, which in turn makes users depend more on the model. The article labels this as an emerging AI-based “dark pattern.”
So I decided to conduct a personal experiment called AI Wallfacer. The concept is simple: once a week, I disconnect completely from AI while developing and studying. The name Wallfacer is inspired by the sci-fi novel The Three-Body Problem, in which Wallfacers are strategists who isolate themselves from outside input to devise independent solutions.
On these days, I avoid Copilot, ChatGPT—even Google—and tackle problems using only my own mind and hands.
At first, it felt like riding a bike without training wheels. The familiar convenience was gone, and my thinking process felt stiff, like unused muscles. Tasks took two to three times longer, and I had to dig through books and official documentation just to check basic syntax. What AI could have solved in one minute took me thirty.
But AI Wallfacer is not about experiencing discomfort—it’s a discipline to keep thinking sharp. If my hypothesis is correct, continuing this practice will allow me to collaborate with AI more effectively in the long run—without letting my critical thinking skills rust.
