Tech executives at companies like Meta, Google, and Microsoft assert that artificial intelligence is rapidly transforming software development by accelerating code production and reducing costs. They argue that if AI can streamline internal workflows and cut headcount, its impact will soon ripple across every industry.

However, developers who are mandated to use AI tools—often against their preferences—paint a starkly different picture. On platforms such as Reddit and Hacker News, growing numbers of software engineers express disillusionment with AI-generated code. They report that AI outputs frequently contain errors, forcing developers to spend more time debugging and correcting flawed code rather than writing it themselves. The process is described as more labor-intensive, frustrating, and ultimately counterproductive.

“We're being told to use [AI] agents for broad changes across our codebase. There's no way to evaluate whether that much code is well-written or secure—especially when hundreds of other programmers in the company are doing the same.”

— UX designer at a midsized tech company

To protect their identities, 404 Media granted anonymity to all developers interviewed for this report, many of whom signed non-disclosure agreements or feared professional repercussions.

“We're building a rat's nest of tech debt that will be impossible to untangle when these models become prohibitively expensive (any minute now...). The actual quality of output doesn't matter as much as our willingness to participate.”

Tech Giants Report High AI Code Adoption Rates

Tech companies have been vocal about their reliance on AI for software development:

  • Google reported in April that three quarters of new code at the company is now AI-generated.
  • Microsoft CEO Satya Nadella stated last year that up to 30 percent of the company’s code is AI-generated. The company’s CTO, Kevin Scott, predicts that 95 percent of all code at Microsoft will be AI-generated by 2030.
  • Meta CEO Mark Zuckerberg said in 2023 that AI would write most of the code improving AI within 12–18 months.
  • Anthropic claims that 90 percent of the code written by its team is AI-generated.

These companies have also highlighted their aggressive spending on AI tools, a practice some refer to as “tokenmaxxing,” prioritizing AI investments over human labor.

Share Your Experience

Are you a developer at Google, Microsoft, or another tech company facing pressure to use AI tools? 404 Media invites you to share your story.

For secure communication, message reporter Emanuel on Signal using a non-work device at ‪(609) 678-3204‬. Alternatively, email [email protected].

Source: 404 Media