---
aliases:
  en:
  - FAI
  - Friendly Artificial Intelligence
concept_id: friendly_ai
preferred:
  en: Friendly AI
  ja: フレンドリーAI
relations:
  relations:
    prerequisite:
    - ai_alignment
    - human_values
status: active
---

# Friendly AI

An older term, especially associated with Eliezer Yudkowsky and MIRI, for an advanced AI that reliably produces beneficial outcomes by being aligned with humane or otherwise appropriate values; the broader research area is now more commonly discussed under AI alignment.

## prerequisite

- [AI alignment](../terminology/ai_alignment.txt)
- [human values](../terminology/human_values.txt)

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