基礎

Constitutional AI

定義

A training methodology developed by Anthropic in which an AI model is guided by a written set of principles (a "constitution") to self-critique and revise its outputs. Constitutional AI is one approach to building safer, more controllable AI systems at scale.

関連用語

Reinforcement Learning from Human Feedback (RLHF)

A training technique that refines language model behaviour by learning from human preferences rather than fixed labels. RLHF is a primary method used to align LLMs like ChatGPT and Claude with desired values and reduce harmful outputs.

AI Alignment

The research and engineering discipline of ensuring that AI systems pursue goals and exhibit behaviours that match human intentions and values. Misalignment risks range from models following instructions too literally to more speculative long-term risks discussed in AI safety literature.

AI Safety

A multi-disciplinary research field concerned with preventing AI systems from causing unintended harm—including technical failures, misuse, and long-term societal risks. In the EU, AI safety is operationalised through the AI Act's risk classification and conformity assessment requirements.

AIの理解にお困りですか？

Physical AI 適合性コールをご予約いただき、これらのAI概念が貴社の業界や課題にどう適用されるかをご相談ください。