Review time is up 441% under vibe coding. That is the real problem.
A state-of-the-art review of the first body of evidence on AI-assisted development was posted to arXiv on 20 Aug 2026. It is a preprint, not itself peer reviewed, though the field experiments it surveys are (Michels et al., arXiv:2608.20446). The number that matters for review is not in any vendor deck: team-level telemetry shows code-review time went up 441% under what the authors call vibe…
A recent review of AI-assisted development, published on August 20, 2026, on the preprint server arXiv, has raised concerns about the impact of "vibe coding" on code review time. Vibe coding is a workflow where developers describe intent and validate by running rather than reading the generated code. The review finds that team-level telemetry shows code-review time has increased by 441% under this method.
However, productivity records are contradictory, with peer-reviewed field experiments reporting a 26% increase in tasks per week and independent randomized trials measuring a 19% slowdown. The authors argue that both results can be true if measurement method, scope, and horizon are held constant. The review highlights that output volume is often conflated with productivity, and self-reported data diverges from independent measurements.
The authors propose that the gains from AI-assisted generation are real for new code but shrink or reverse on mature codebases. The review does not settle on specific tools to fix review throughput, as no single tool is the focus of the study. While the review reports reliable code generation, it also documents weak fault detection, weak documentation, code-quality degradation, and security failures in deployed applications.
The 441% figure is a useful anchor for the review-load conversation, and the direction of the problem is documented across multiple independent studies.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
