---
aliases:
  en:
  - misaligned artificial intelligence
  ja:
  - ミスアラインド
concept_id: misaligned_ai
preferred:
  en: misaligned AI
  ja: ミスアラインドAI
relations:
  relations:
    broader:
    - artificial_intelligence
    prerequisite:
    - artificial_intelligence
    - misalignment
status: active
---

# misaligned AI

An AI system whose goals, optimization, or behavior are materially misaligned with the human values, interests, intentions, or other target objectives relevant to its design or deployment.

## broader

- [artificial intelligence](../terminology/artificial_intelligence.txt)

## prerequisite

- [artificial intelligence](../terminology/artificial_intelligence.txt)
- [misalignment](../terminology/misalignment.txt)

## Documents: mentions

- [アラインメント研究の自動化は、研究AIが意図的に妨害しなくても失敗し得る](../documents/bowkis_et_al.automated-alignment-can-fail-without-deliberate-sabotage.txt)
- [現在と大きく変わらない条件で人間を上回るAIを作れば、人類絶滅が最も可能性の高い結果となる](../documents/yudkowsky.superhuman-ai-under-current-conditions-would-likely-cause-extinction.txt)

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