Software Developers Say Heavy AI Use Is Making Them Forget How to Code

Software Developers Face Mandatory AI Adoption at Work

Software developers at major technology companies are sounding alarms about how AI in coding is fundamentally changing their profession—and not for better. According to interviews conducted by 404 Media, programmers are either explicitly ordered to use AI tools or find that heavy AI use factors directly into their performance reviews 3

3

. At one FAANG company, a software engineer revealed that "AI in some shape or form is all but explicitly mandated" and that "its usage is part of our performance review criteria" 3

3

. This pressure has led to what developers call "performative" AI use, where the actual quality of output matters less than demonstrating willingness to participate in the AI transformation.

The scale of AI-generated code deployment is staggering. Google announced in April that three quarters of new code at the company is now AI-generated, while Anthropic reports that 90 percent of code written by most of its team comes from AI 3

3

. Microsoft CEO Satya Nadella said up to 30 percent of the company's code was generated by AI, with the company's CTO predicting that figure will reach 95 percent by 2030 3

3

. Yet this productivity surge hasn't translated into better products or shorter work weeks—instead, it has primarily justified massive layoffs, with Meta cutting 10 percent of its workforce (around 8,000 people) and Microsoft offering voluntary retirement to 7 percent of its American workforce (around 125,000 people) 3

3

.

Programmers Skills Are Deteriorating Through De-Skilling

The consequences of vibe coding extend beyond workplace dynamics to affect developers' core competencies. One engineer told 404 Media about a frightening moment when they forgot how to implement a Laravel API—a framework they had studied in university and used professionally for years. "It feels like I am back before I ever wrote a single line of code," the developer confessed 4

4

. This phenomenon of losing the ability to code mirrors what happened when people stopped memorizing phone numbers after smartphones became ubiquitous 1

1

.

Another software engineer described the cognitive decline more bluntly: "It's making me dumber for sure. It's like when we got cellphones and stopped remembering phone numbers, but it's grown to me mentally outsourcing 'thinking' in general" 4

4

. The developer noted that critical thinking skills and the ability to reason about problems have degraded because generative AI is always just a question away. A UX designer at a midsized tech company expressed concern about being told to use AI agents for broad changes across codebases, noting there's "no way to evaluate whether that much code is well-written or secure" 3

3

.

AI Is Rotting Their Brains and Creating Technical Debt

The sheer volume of AI-generated code creates two critical problems that threaten long-term productivity. First, generative AI produces too much code for humans to conduct effective code audits and debugging 2

2

. Developers either grow exhausted from analyzing AI output or cannot understand it at all, ultimately shipping products without knowing whether they contain flaws. One programmer warned that companies are "building a rat's nest of tech debt that will be impossible to untangle" when AI models become prohibitively expensive 3

3

.

This bad-code debt will likely surface only when these codebases require updating or adjusting in the future—by which point no one will understand how the systems work because they "vibe-coded their way to the initial version" 1

1

. Some developers worry that while experienced coders might adapt, vibe coding could endanger the next generation's ability to analyze code or perform basic tasks 2

2

. The promise that AI would handle boring daily tasks and free humans for creative work appears to be a pipe dream .

The Return to Manual Coding Gains Momentum

Across Reddit, Hacker News, and other developer communities, more programmers are actively seeking to return to manual coding 1

1

. Commentators on Reddit explain that even though coding by hand takes longer, it provides deeper understanding of the code, enabling more effective debugging 2

2

. One developer who pledged to abandon AI coding tools wrote that while AI can effectively write code for individual features, it cannot handle architecture and loses context when projects grow beyond a certain size 2

2

.

Some programmers see additional challenges ahead. As companies like Microsoft switch to per-token billing, concern is mounting that AI pricing will rise dramatically in the near future, prompting companies to use AI less while making it harder to fix the mess AI has created 1

1

. For developers trying to preserve their abilities despite workplace mandates, the stakes are clear: without raw coding skills, they're just shepherding error-prone AI 1

1

. The atrophying skills phenomenon raises questions about whether the tech industry can find a sustainable balance between AI assistance and human expertise, or whether the current trajectory will leave organizations dependent on tools that undermine the very skills needed to maintain complex systems. A growing body of research supports the idea that critical thinking skills are being diminished by AI use, with burnout increasing among developers who must constantly prompt AI while reviewing its output 4

4

.