Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I had a couple of great examples of getting the exact opposite answer depending on how I worded my question, now I can't trust any of the answers.


Maybe the answer to your question was subjective?


The first set of questions was about code review timing in a workflow (background: was in a discussion about different styles of workflows):

Q1.1: "do most developers do code reviews before testing"

A1.1: essentially "yes most before..."

Q1.2: "do most developers do code reviews after testing"

A1.2: answer was essentially "yes most after..."

The 2nd set of questions was related to building a retaining wall. I've never used the type of wall block with a center notch and groove, only the type with a lip on the back. The center type creates a wall straight up but lip in back leans back a little. I was curious if one was more stable than the other:

Q2.1: "is retaining wall block with center groove more stable"

A2.1: essentially "yes, block with center groove is more stable..."

Q2.2: "is retaining wall block with back lip more stable..."

A2.2: essentially "yes, block with lip on back is more stable..."

When I finally found an engineering website with real details, the answer was that the lip on back was a little more stable due to the resulting angle of the wall. It also emphasized that the lip and the center groove are not factored in to stability calcs at all, they are only for alignment, gravity and friction of the blocks is what holds it in place.


It makes me wonder, why don't we want AI to be deterministic and make it nondeterministic on purpose instead? I hope someone can explain it to me...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: