Generation Is Cheap, Review Is Expensive: How to Stop Shipping AI Slop — Gabriel Martinez, G2i

Reframes AI code quality as a review-capacity problem and offers practical countermeasures: small reviewable changes, diagrams as the review surface, and clear ownership.

• Martinez defines slop not as AI-written code but as work that looks finished before the thinking is: ambiguity cleared with guesses, with nobody noticing there was a choice to make.
• He argues code is still a liability, agents make a messy codebase messier faster, and quick prototypes convince stakeholders the hard part is done.
• He flags thousand-line PRs and 20-page AI-written docs as ways of pushing the thinking onto reviewers. His fixes borrow from Rails: strong conventions alongside small changes and clear ownership.

• videoWhy Your Company Needs a Context Graph (and How to Build It) — Gil Feig, Merge
• videoSmall Models, Big Results: Training a Finance Agent for Under $500 — Charles Dickens, Snorkel AI
• videoFrom Context to Memory: Your Agents Need a Real Memory Layer — Anders Swanson, Oracle
• videoWhy 80% Reliability Isn't Good Enough — Felipe Blanes, Amazon AGI Lab

• videoThe Death of the Code Review: What the Data Actually Says — Laurie Voss, Arize AIAI Engineer
• videoThe Last Human Code Review: Building Trust in AI-Generated Code — Itamar Friedman, QodoAI Engineer
• videoAI Writes More PRs. Who Validates Them? — Ali-Reza Adl-Tabatabai, SonarAI Engineer

Checking sign-in…

Loading comments…