A software engineer with three years of experience has admitted to not understanding their own code anymore after becoming dependent on artificial intelligence (AI).
In a social media post, the techie said they joined the corporate world straight out of college around the time Large Language Models (LLMs) were becoming popular worldwide.
A software engineer admits AI reliance reduced code understanding and critical thinking.
If you took away AI tomorrow, I feel like I’d code at the level of a first-semester freshman. It started with fixing smaller bugs or basic features, where they could follow the logic but as the magnitude of tasks grew, they started falling behind in understanding. “All I do is feed requirements into Claude, accept what it spits out, and push. Half the time, I don’t even fully understand the architecture or the code it just generated,” the techie wrote. “The bitter irony? “I feel completely useless, fraudulent, and terrifyingly replaceable. Am I cooked? Has anyone else dug themselves into this hole and actually managed to climb back out?”
Because my brain feels fried, i’m even using AI to clean up this post.
When seniors ask questions about the code, the techie said their ‘stomach drops’ as they cannot defend or explain the implementation without feeding their comments back into an LLM. My focus is shot, my critical thinking feels atrophied, and I find myself unable to sit down and read documentation or trace code manually anymore,” they said.
Since then, they had become reliant on AI for ‘pretty much everything’.
Pressure deadlines are as such that there’s little time to understand things. Meanwhile, another added: “Started happening to me also. I scaled back the AI usage. I let it code and write tests, but I verify the code before pushing and creating a PR. Dev testing is always manual with debugging, so I know what the code does. A third commented: “Your job is not to write code but to deliver features/products a customer is willing to pay for. That has always been the case. Eventually the only metric that matters is the money you save or bring in. In the last three months, I was contracted at four companies to save their tech products. The issues were the same.
Because nobody had any clue what they were writing.”, hundreds of bugs & regressions.
As the post gained traction, one of the social media users said: “Very normal nowadays. A fourth said: “Cognitive debt is one of the biggest risks with overuse of LLMs.
