AI Agents Created AI Code Slop, But Can They Fix It?

AI coding agents may have streamlined coding pipelines but they’ve also created a serious AI code slop challenge. Back in September, agent code review startup Qodo released The State of AI Code Quality report, which surveyed 500 U.S. software developers and 300 U.S. engineering leaders and found that 89% of organizations report having had an AI-related production incident.

The study found that just 3.7% of engineering leaders say their existing processes are sufficient to maintain quality and governance as agents take on more work. At the same time, 36.4% of developers say reviewing AI-generated code requires more cognitive effort.

Across the software engineering industry, developers are grappling with managing AI slop code, not only experimenting with using agents to generate code at scale, but also to review code too.

Inside AI Code Slop

A number of solution providers have been aiming to address the growth in AI slop. Code review platform Gitar, founded in 2023 by ex Uber engineers Ali-Reza Adl-Tabatabai and Gautam Korlam, was acquired by AI code verification provider Sonar Source in May, and uses a code verification agent to flag issues in AI-generated code.

Gitar cofounder and now EVP of transformation at Sonar, Ali-Reza Adl-Tabatabai told me in a video interview that the “code slop issue is real," which can be defined as code that introduces bugs or vulnerabilities.

“There’s a lot of various forms of slop,” Adl-Tabatabai said, highlighting that AI tends to generate large code changes, introducing unnecessary abstractions and making code more complicated than it needs to be. He also added that unless you take actions to try to prevent slop from getting into code, it can propagate and cause other problems.

“These are things that you know if they make their way into your codebase, they could cause you trouble down the line in terms of issues in production, or generally just make your code harder to maintain and less efficient," Adl-Tabatabai said.