← News

ResearchSources: arXiv · 18 Nov 2026

Systematic review finds AI coding assistant productivity evidence fragmented and hard to compare

A staged systematic literature review of AI coding assistants reports 187 distinct productivity markers across studies, with positive effects reported more often in perceived findings than in objectively measured ones.

In plain words

Researchers reviewed studies on whether AI coding assistants make developers more productive. They found the research is hard to compare because each study measures productivity differently, and positive results appear more often when developers self-report than when effects are measured objectively.

What was announced

The review, published as an arXiv preprint, argues that the central problem in this literature is not a shortage of studies but a lack of comparability: prior work measures productivity through heterogeneous markers, settings, and reporting styles, which makes cross-study conclusions difficult. In a first stage the authors synthesized studies that explicitly examine developer productivity; in a second stage they widened the lens to productivity-related developer outcomes using the SPACE framework, which captures human-centered evidence that explicit productivity studies often omit.

Across both stages the authors identified 187 unique productivity and productivity-related markers, while individual studies typically used only five, which the review characterizes as a broad but highly fragmented evidence base. Explicit productivity research leaned on time-, output-, and artifact-oriented proxies, whereas the SPACE-aligned corpus gave substantially more visibility to satisfaction, psychological aspects, and well-being. Reported effects were often positive, but the authors note this pattern is stronger in perceived and tendency-based findings than in objective or statistically evaluated ones. They conclude that AI coding assistants can support productivity, but that current evidence does not yet support simple, context-independent conclusions.

What it does not show

  • The source is the paper's abstract only; the full review's methods, study selection, and detailed findings are not available in the supplied material.
  • The review's conclusions rest on the authors' own synthesis; no independent confirmation of the findings is provided.

Sources

Source-support review completed 19 Nov 2026. This story was drafted with AI assistance and checked against retained source passages. Automated checking is not independent verification.

Coverage: arXiv: abstract.

Updates and corrections

No corrections so far.

First published 19 Nov 2026

New to the subject? AI from the Ground Up starts from the beginning.