---
alternate_urls:
- https://www.lesswrong.com/posts/GNhMPAWcfBCASy8e6/a-central-ai-alignment-problem-capabilities-generalization
authors:
- Nate Soares
citation:
  locator_schemes:
  - anchor
id: soares.sharp-left-turn
kind: essay
language: en
published_at: '2022-06-15'
title: 'A central AI alignment problem: capabilities generalization, and the sharp
  left turn'
url: https://www.alignmentforum.org/posts/GNhMPAWcfBCASy8e6/a-central-ai-alignment-problem-capabilities-generalization
---

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

Original: <https://www.alignmentforum.org/posts/GNhMPAWcfBCASy8e6/a-central-ai-alignment-problem-capabilities-generalization>



[Corpus index](../index.txt)