Should engineers read AI-generated code or just check the outputs, according to A Life Engineered's guest?

Asked by Steve Huynh whether he is on team read the code, the guest says yes for his own work, which is mathematical computing and AI internals. There a wrong minus sign or an integer used where a float belongs is invisible in the outputs; even an LLM built from scratch cannot be verified just by looking at what it produces. Among clients, he finds the choice is driven more by personality and the kind of work than by engineering management.

• The guest knows heavy AI users who still read all of a 10,000-line generated change, which is slower but still faster than writing it by hand.
• Other engineers he respects generate artifacts and judge them only by whether the test suite passes.

▸In their words

“For me personally, I'm on team read the code.” Guest · at 4:49 — open in the transcript →

Source: our own transcription of the episode. The full conversation is on A Life Engineered (2026-10-27). Read the whole episode summary →

▸More from this episode