AI ALIGNMENT FORUM
AF

Sharp Left Turn

Edited by Multicore, et al. last updated 30th Dec 2024

Sharp Left Turn is a scenario where, as an AI trains, its capabilities generalize across many domains while the alignment properties that held at earlier stages fail to generalize to the new domains.

See also: Threat Models, AI Takeoff, AI Risk

2

2

Posts tagged Sharp Left Turn

74A central AI alignment problem: capabilities generalization, and the sharp left turn

3y

18

37Refining the Sharp Left Turn threat model, part 1: claims and mechanisms

Vika, Vikrant Varma, Ramana Kumar, Mary Phuong

3y

3

16We may be able to see sharp left turns coming

Ethan Perez, Neel Nanda

3y

25

86“Sharp Left Turn” discourse: An opinionated review

7mo

13

23We don't understand what happened with culture enough

2y

3

30Reframing inner alignment

3y

9

19Refining the Sharp Left Turn threat model, part 2: applying alignment techniques

Vika, Vikrant Varma, Ramana Kumar, Rohin Shah

3y

5

67Evolution provides no evidence for the sharp left turn

2y

12

46Response to Quintin Pope's Evolution Provides No Evidence For the Sharp Left Turn

2y

0

23Smoke without fire is scary

3y

9

22It matters when the first sharp left turn happens

3y

1

9Disentangling inner alignment failures

3y

4

Add Posts