Do humans still need to review every line of AI-generated code?
On October 30, Ruby on Rails creator David Heinemeier Hansson posted that believing humans still need to review every line of code is the real "AI psychosis." The post drew hundreds of replies and quote posts from developers debating how much human review AI-written code requires.
Side A
Line-by-line human review is outdated
This camp argues that insisting on human review of every line of AI-written code is a mindset developers should move past. Hansson wrote that treating AI as just another tool or requiring humans to review every line is the real AI psychosis, and separately said AI is even better at reviewing code than writing it. Other developers in this camp said line-by-line review should give way to tests, verification and automated checks, and that being useful now means building those checks rather than reading every line.
Evidence posts
Overlap
Where they overlap
No clear common ground was found in the posts reviewed.
Side B
Human review is still needed
This camp argues that humans still need to read AI-written changes before they reach production. Developers in this camp said that when systems break overnight, a person rather than the model is accountable, so they still read diffs before merging. One cited a reported scale-back of an agent-heavy plan at Meta after incidents rose, and said somebody has to read code before production. Computer scientist Santiago Valdarrama questioned whether time saved by agent reviews is lost to fixing bugs afterward.
Evidence posts
Reader vote
Be the first to weigh inWhere do you land? Sign in to vote — one account, one vote.
Free newsletter
Both Sides Daily
One email a day with the day's debates: what each side says, where they meet, and how readers voted. No spam, unsubscribe with one click.
This page was compiled by AI and published after automated checks. Something wrong?Report a correction