Reproduce “Conversational Planning for Personal Plans” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Konstantina Christakopoulou, Iris Qu, John Canny, Andrew Goodridge, Cj Adams, Minmin Chen, Maja Matarić
Researchers from Google DeepMind and Google developed a hierarchical AI framework enabling conversational planning for complex, long-term personal goals. The system generates, refines, and adapts personalized plans through natural language interaction, demonstrating its capabilities in applications like learning assistance and fitness coaching by providing relevant content and incorporating user feedback.
Key insights
- A hierarchical framework, inspired by hierarchical reinforcement learning, allows AI systems to reason at different levels of abstraction, from high-level planning to specific implementation details.
- Large Language Models (LLMs) can serve as meta-controllers for high-level decision-making and specialized sub-policies for executing specific macro-actions in a conversational planning system.
- Integrating natural language feedback and proactive questioning allows for dynamic plan refinement and personalization, closely mirroring human-like assistance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.