Lo scenario
A user interacts with an AI assistant to accomplish a task while ensuring the AI's actions remain aligned with user goals and values through iterative feedback and model updates.
Cosa contiene questo disegno
Leggi le decisioni che ne stanno alla base.
01
AI asks clarifying questions when intent is ambiguous
02
AI revises proposal when alignment check fails
03
User provides corrective feedback to update value model
Reusable for designing human-in-the-loop AI systems, chatbots, or any interactive AI requiring continuous alignment with user preferences.