CoLa: Learning to Interactively Collaborate with Large Language Models
This work addresses the challenge of enhancing human-AI collaboration by automating guidance strategies, offering a novel approach for scalable problem-solving in language tasks.
The paper tackles the problem of simulating human guides to collaborate with large language models (LLMs) for complex language tasks, introducing CoLa, a self-guided learning paradigm that outperforms competitive approaches across multiple domains, with a small-sized trained guide surpassing GPT-4 in guiding performance.
LLMs' remarkable ability to tackle a wide range of language tasks opened new opportunities for collaborative human-AI problem solving. LLMs can amplify human capabilities by applying their intuitions and reasoning strategies at scale. We explore whether human guides can be simulated, by generalizing from human demonstrations of guiding an AI system to solve complex language problems. We introduce CoLa, a novel self-guided learning paradigm for training automated $\textit{guides}$ and evaluate it on two QA datasets, a puzzle-solving task, and a constrained text generation task. Our empirical results show that CoLa consistently outperforms competitive approaches across all domains. Moreover, a small-sized trained guide outperforms a strong model like GPT-4 when acting as a guide. We compare the strategies employed by humans and automated guides by conducting a human study on a QA dataset. We show that automated guides outperform humans by adapting their strategies to reasoners' capabilities and conduct qualitative analyses highlighting distinct differences in guiding strategies.