The age of humans reviewing code directly is largely over, argues @0xdoug
StefanoGogioso · x · 2026-10-25
A widely circulated take: humans should still review software, but the right way is to converse with the LLM about the code rather than read it line by line. The argument: our brains are bad at math/code reasoning; natural language is the mind's native protocol, and modern LLMs make far fewer mistakes translating code to/from language than humans make lowering semantic reasoning into code. The quoted tweet adds field evidence: Opus 5.5 writes better code than humans given right direction and context—PR reviews are now about high-level architecture and direction.
More from AGI Musings
- Musk softens on AI risk: 'maybe Larry Page was right,' hopes AI is nice to us — marigo · 2026-10-25
- With AI Shipping in Days Not Months, Practitioners Say Speed Should Raise Quality Bars — annetgriffin · 2026-10-25
- Ex-OpenAI researcher: the 'attacker diffusion lag' shielding us from AI misuse may close fast — joshua_saxe · 2026-10-25
- Samotsvety forecasters put max unemployment by 2030 at ~9.7% aggregate (7-9% typical) — NathanpmYoung · 2026-10-25
- Median Americans Think the Typical Corporation Makes 36% Net Profit — NathanpmYoung · 2026-10-25
- Sam Altman: 'I no longer think we're gonna be able to just patch all the bugs' — GarrisonLovely · 2026-10-25