---
aliases:
  en:
  - prosaic AI alignment
concept_id: prosaic_alignment
preferred:
  en: prosaic alignment
  ja: プロセイック・アラインメント
relations:
  relations:
    broader:
    - ai_alignment
    prerequisite:
    - ai_alignment
    - machine_learning
status: active
---

# prosaic alignment

AI alignment research aimed at powerful or general AI that arises by scaling and extending broadly contemporary machine-learning methods without requiring fundamentally new ideas about how intelligence works. The term does not mean such systems are aligned by default; it names the problem of aligning this comparatively 'prosaic' route to advanced AI.

## broader

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

## prerequisite

- [AI alignment](../terminology/ai_alignment.txt)
- [machine learning](../terminology/machine_learning.txt)

## Documents: about

- [既に形成された欺瞞的な方策は、標準的な安全性訓練を経ても残り得る](../documents/hubinger_et_al.deceptive-policies-can-persist-through-safety-training.txt)
- [チャット形式の安全性訓練で望ましく振る舞うことは、エージェント的な状況での安全性を保証しない](../documents/macdiarmid_et_al.chat-alignment-does-not-establish-agentic-alignment.txt)
- [行動訓練で望ましく振る舞っても、意図した目標の頑健な汎化は保証されない](../documents/ngo_et_al.behavioral-training-does-not-establish-robust-goal-generalization.txt)

## Documents: mentions

- [狭いタスクでファインチューニングしても、影響がそのタスク内に限られるとは限らない](../documents/betley_et_al.narrow-finetuning-can-produce-broad-misalignment.txt)
- [訓練中に従うようになっただけでは、競合する選好が置き換わったとは言えない](../documents/greenblatt_et_al.training-compliance-does-not-establish-preference-change.txt)

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