SCOOP: A Framework for Proactive Collaboration and Social Continual Learning through Natural Language Interaction andCausal Reasoning

TL;DR

The paper addresses the challenge of AI assistants operating in dynamic, multimodal environments where users' goals and beliefs are not fully accessible. It proposes SCOOP, a social continual learning framework that enables proactive collaboration through natural language interaction and causal reasoning, featuring two architectures: a Base Oracle-Aided ReAct and an Advanced ReAct with a causal world model and a causal knowledge graph. The approach combines dialogue-driven knowledge acquisition, question generation, and planning under uncertainty, using an external oracle and causal-inference libraries to refine causal graphs and guide decision-making. This framework aims to improve robust inference and adaptability under resource/interaction constraints, with evaluation paradigms inspired by developmental psychology to assess knowledge-gap identification, questioning quality, and amortization of knowledge costs across tasks.

Abstract

Multimodal information-gathering settings, where users collaborate with AI in dynamic environments, are increasingly common. These involve complex processes with textual and multimodal interactions, often requiring additional structural information via cost-incurring requests. AI helpers lack access to users' true goals, beliefs, and preferences and struggle to integrate diverse information effectively. We propose a social continual learning framework for causal knowledge acquisition and collaborative decision-making. It focuses on autonomous agents learning through dialogues, question-asking, and interaction in open, partially observable environments. A key component is a natural language oracle that answers the agent's queries about environmental mechanisms and states, refining causal understanding while balancing exploration or learning, and exploitation or knowledge use. Evaluation tasks inspired by developmental psychology emphasize causal reasoning and question-asking skills. They complement benchmarks by assessing the agent's ability to identify knowledge gaps, generate meaningful queries, and incrementally update reasoning. The framework also evaluates how knowledge acquisition costs are amortized across tasks within the same environment. We propose two architectures: 1) a system combining Large Language Models (LLMs) with the ReAct framework and question-generation, and 2) an advanced system with a causal world model, symbolic, graph-based, or subsymbolic, for reasoning and decision-making. The latter builds a causal knowledge graph for efficient inference and adaptability under constraints. Challenges include integrating causal reasoning into ReAct and optimizing exploration and question-asking in error-prone scenarios. Beyond applications, this framework models developmental processes combining causal reasoning, question generation, and social learning.