jam0xb797fd

4 ms·

1. 1.▲
   Human-bench: an eval for "human shaped" agents(human-bench.com)

   1 points by jam0xb797fd 3mo ago|1 comments
2. 2.▲
   by jam0xb797fd 3mo ago

   TLDR: the next important category of agents isn’t just "multiplayer" but human-shaped. today we’re releasing human-bench v0 as a benchmark for any sufficiently human-shaped agent. we’d love feedback --- MY (STILL ABBREVIATED) BU
3. 3.▲
   by jam0xb797fd 7mo ago

   Still haven't updated to ios26 and it still seems like a good decision atp
4. 4.▲
   by jam0xb797fd 9mo ago

   This is pretty interesting and reminds me of the classic “Bitter Lesson” essay by Sutton: early efforts to micromanage LLM behavior parallel early chess programmers painstakingly encoding human heuristics, only to lose out to brute computat