One Thousand Faces

Post Truth

The inevitable breakdown of consensus reality, and why that's a ~good thing

Oct 6 • 

Hero Thousandfaces

12

August 2025

WikiBench: 76% of SOTA Models Fail

Assessing real-world understanding and agentic tool use with a twist on a classic game.

Aug 1 • 

Hero Thousandfaces

30

5

June 2025

How To Be Funny, Part 2: For AIs and The People That Love Them

So, deep learning walks into a wall... (AKA my gripes with current post-training and benchmark paradigms)

Jun 16

10

March 2025

How To Be Funny, Part 1: For Humans

If you've never gotten off a good joke in your life, start here

Mar 12 • 

Hero Thousandfaces

20

1

January 2025