Strange Loop Canon

Seeing like an agent

AI agents as digital daemons

Dec 8 • Rohit Krishnan

57

11

13

Contra Scott on AI safety and the race with China

Dec 2 • Rohit Krishnan

26

17

3

November 2025

Epicycles All The Way Down

“All models are wrong, but some are useful.” — George E.

Nov 26 • Rohit Krishnan

50

18

11

October 2025

Poisoned prose

on semantic trojan horses in LLMs

Oct 27 • Rohit Krishnan

31

19

4

Can we get an AI to write better?

A small step

Oct 6 • Rohit Krishnan

48

36

10

September 2025

Anomie

the vibes they are a-changin

Sep 28 • Rohit Krishnan

45

9

6

Prediction is hard, especially about the future

more reinforcement learning, this time on the future

Sep 16 • Rohit Krishnan

46

10

4

August 2025

The future of work is playing a videogame

Aug 25 • Rohit Krishnan

72

15

4

Walter

experiments in rlnvr

Aug 23 • Rohit Krishnan

54

5

7

July 2025

Ads are inevitable in AI, and that's okay

Convergent evolution in LLMs will get us there

Jul 28 • Rohit Krishnan

37

21

7

The Slow Apocalypse: When will we run out of kids?

More than you wanted to know about the fertility crisis

Jul 16 • Rohit Krishnan

32

17

7

Seeing like an LLM

"I will run the tests again. I expect nothing. I am a leaf on the wind." an LLM while coding

Jul 9 • Rohit Krishnan

45

9

12