News for hackers. Stuff that matters.

A comment on Some uncomfortable truths about AI coding agents. Parent comment

Re:Some uncomfortable truths about AI coding agents by raw_anon_1111 on Saturday March 28, 2026 @05:47PM (#47558405)
Sort of a devils advocate question. If you write and review your tests and the functional and non functional requirements and the human tests for usability pass, why does the code matter?

Non functional requirements: performance, security, reliability, logging etc?

[ Parent | on HN ]
Re:Some uncomfortable truths about AI coding agents by catgary on Sunday March 29, 2026 @07:28PM (#47568563)
Because the code is the actual thing, tests can only show that the code fails in certain cases, they don’t actually prove the code is correct.
[ Parent | on HN ]
Re:Some uncomfortable truths about AI coding agents by raw_anon_1111 on Sunday March 29, 2026 @09:59PM (#47569585)
If you are writing the correct tests that mirror the requirements, why wouldn’t passing tests mean the code is correct?
[ Parent | on HN ]
Re:Some uncomfortable truths about AI coding agents by catgary on Wednesday April 01, 2026 @10:33AM (#47601506)
Because it doesn’t? Thats why the field of formal methods exists.
[ Parent | on HN ]
Re:Some uncomfortable truths about AI coding agents by raw_anon_1111 on Wednesday April 01, 2026 @12:39PM (#47603216)
Then that sounds like you aren’t writing good tests…
[ Parent | on HN ]