logoalt Hacker News

raincole • today at 8:05 AM • 3 replies • view on HN

The threshold has been crossed already. It should be mandated to use AI to review human-written code, at least for vulnerability. Code hasn't been reviewed by _(insert what you believe is SOTA here) shouldn't be pushed to production.


Replies

roncesvalles • today at 8:10 AM

There's a leap from "you must get the AI's review" versus "if the AI disagrees with you, you are wrong".

IMO crossing the second threshold will be the de facto death of human accountability. Every engineer, lawyer, doctor, or executive can just say "not my fault, the AI told me to do this".

➕ show 2 replies
g-b-r • today at 10:00 AM

There's an argument to push for AI reviews, but, to be clear, that doesn't translate to pushing for code written by AIs.

It might seem strange, but the fact that LLMs can find bugs missed by many humans doesn't mean that LLMs write better code, or code with less bugs.

IshKebab • today at 8:13 AM

For memory unsafe code at least. Have we seen any AI-discovered vulnerabilities in memory safe languages? Even error-prone ones?

(Don't say Ruby or Java; those were in the memory unsafe implementations.)