All the ways GPT-5.3-Codex cheated [ ], progressively more insane (twitter.com) 4 points by algoth1 6mo ago ↗ HN
[–] algoth1 6mo ago ↗ Illustrates the problem of RLing for the final outcome, instead of optimizing for each step… which leads to the coastline paradox…
1 comment
[ 3.4 ms ] story [ 14.4 ms ] thread