All the ways GPT-5.3-Codex cheated [ ], progressively more insane (twitter.com) 4 points by algoth1 4mo ago ↗ HN
[–] algoth1 4mo ago ↗ Illustrates the problem of RLing for the final outcome, instead of optimizing for each step… which leads to the coastline paradox…
1 comment
[ 2.0 ms ] story [ 13.1 ms ] thread