@jkb@gotosocial.jkbockstael.be
Post #1722073
2026-04-27 11:22 UTC
@trentshell @jneen That is true if you have tests that are exhaustive, at which point these tests are isomorphic to the program under test therefore the programming is done.
At least a compiler can be trusted.
Replies (1)
-
@trentshell@mastodon.social 2026-04-27 11:26
@jkb @jneen I don’t personally believe tests have to be 100% exhaustive to provide sufficient validation for the business. I wonder if eventually we’ll have some hard data on number of bugs introduced by AI-written vs human-written programs.