Coders are refusing to work without AI — and that could come back to bite them

2026-05-29 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Intermediate, short

Summary

By 2026, developers are largely unwilling to work without AI coding tools, a trend highlighted by METR research. While a 2025 study initially found AI slowed developers due to error correction and steering, a May 2026 METR survey indicated developers perceive themselves as twice as productive. This perceived productivity is questionable. "Tokenmaxxing" led Amazon to shut down its Kirorank leaderboard due to excessive AI agent use and cost overruns. Uber also exhausted its 2026 AI budget in four months without measurable productivity gains. AI-generated code may increase maintenance burdens; Entelligence AI claims 44% of tokens are spent on AI-generated bug fixes, and CodeRabbit reported AI code had 1.7x more problems than human code. Singapore Management University researchers warned in April 2026 about long-term maintenance costs. Solutions range from using AI agents for fixes, though current agents like Devin are rated junior/mid-level, to a human-centric approach emphasizing strong quality assurance and human oversight for critical tasks.

Key takeaway

For AI Engineers or Directors of AI/ML evaluating coding tool ROI, recognize that perceived productivity gains from AI may be misleading. Your teams could be incurring significant hidden costs and future maintenance debt, as seen with "tokenmaxxing" and increased bug fixes. Implement robust AI-specific quality assurance and mandate thorough human review of all AI-generated code. Prioritize human expertise for critical tasks like software architecture and security design to mitigate long-term risks.

Key insights

Widespread AI coding tool adoption masks dubious productivity gains and potential increases in maintenance costs.

Principles

• Perceived AI productivity often exceeds actual gains.
• AI-generated code can increase long-term maintenance.
• Human oversight remains crucial for AI-assisted coding.

Method

Programmers should deeply understand AI's task proficiencies, implement strong AI-specific quality assurance systems, and carefully review AI-generated code as if from a junior developer.

In practice

• Implement AI-specific quality assurance.
• Review AI-generated code diligently.
• Prioritize human oversight for architecture.

Topics

• AI Coding Tools
• Developer Productivity
• Code Maintenance
• AI Cost Management
• Quality Assurance
• Software Architecture

Best for: CTO, VP of Engineering/Data, Executive, Software Engineer, AI Engineer, Director of AI/ML

Related on AIssential

• Quoting James Shore — AI coding agents must inversely reduce maintenance costs to justify increased code output.
• Is ‘nearly right’ AI generated code becoming an enterprise business risk? — AI-generated code is rapidly becoming prevalent, but it introduces significant challenges in testing, quality, and security.
• The Hidden Cost of AI-Assisted Programming — AI-assisted programming risks eroding deep engineering intuition, paradoxically increasing the strategic value of profound technical understanding.
• Building Pi, and what makes self-modifying software so fascinating — Human judgment and deliberate friction are crucial for maintaining software quality amidst rapid AI-driven code generation.
• ClawCode: create production grade code with less cleanup debt — AI code generation often creates cleanup debt, requiring a shift from quantity to quality in output.
• Are AI agents actually slowing us down? — Over-reliance on AI coding tools prioritizes speed over quality, leading to increased tech debt and system outages.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.