3 ms·
This is a general problem with trusting computer AI. I am talking about image recognition stuff etc. At the very least we need some invariants on a higher lev
by EGreg 8y ago
This is a general problem with trusting computer AI.
I am talking about image recognition stuff etc.
At the very least we need some invariants on a higher level to kick in and heavily penalize any such results! Not just bottom up but we need top down.
Where is this feeback mechanism?