In the 2024 Accelerate State of DevOps Report, it was found that 76% of developers are using AI tools in their daily tasks, though 40% have little trust in AI.
Some experienced developers feel less productive with AI-generated code, as it can be sloppy, complex, and prone to errors.
AI adoption negatively impacts software delivery performance, affecting both throughput and stability metrics.
AI-generated code often leads to larger pull requests, making reviews more challenging and potentially sacrificing throughput.
Code reviewers may overlook important details in large changes, impacting the efficiency of the review process.
AI-assisted code reviews may not be trusted if the code being reviewed is AI-generated, further complicating the stability of software changes.
Over-reliance on AI for code generation can create an illusion of speed, potentially compromising software quality and stability.
Increased AI adoption may lead to a perceived productivity increase, but could result in poorer quality software and more broken changes.
Vibe coding, dependent on AI for app development, can result in unstable applications, outages, security vulnerabilities, and loss of work.
Software development teams need to be cautious with AI adoption to prevent sacrificing stability for perceived productivity gains.