AI agents made writing code cheap, but reviewing generated code is a different job
alex_verem · x · 2026-10-08
In a widely shared thread, @mardehaym argues that AI coding agents made writing code cheap but didn't make deciding cheap: understanding, questioning, and owning generated code in production still falls on humans. A single PR demands scrutiny the agent never performed — do tests catch real failures, does the abstraction earn its maintenance cost, will schema changes survive years of data, which security assumptions went unchecked. Writing code means understanding the problem as you go; with agents, the answer arrives before you've thought it through, forcing you to reverse-engineer reasoning while more code piles up. The fix isn't faster senior review but converting senior judgment into automated checks.
More from coding & agent
- WSJ columnist declares 'the year of Linux' is finally here, thanks to AI agents — alexvoica · 2026-10-08
- Full materials from a 2-day workshop on using Claude Code and Codex for academic research — JeremyNguyenPhD · 2026-10-08
- Attestation: local-first MCP tools turn research provenance into CI-enforceable claims — JeremyCMorgan · 2026-10-08
- Tensorlake npm SDK 0.5.144 Found Shipping Credential-Stealing Malware — TechNadu · 2026-10-08
- Enterprise agent lesson: permissions don't guarantee correct business transactions — shashib · 2026-10-08
- Meta Data: AI-Reviewed Diffs Have 1/50th the Production-Incident Rate of Human Review — arjunrajlab · 2026-10-08