Quality-Diversity through AI Feedback
This work addresses the problem of limited algorithmic specification for quality and diversity in qualitative domains like creative writing, offering a potentially generalizable approach for AI systems to autonomously innovate, though it appears incremental in applying existing QD methods with AI feedback.
The paper tackles the challenge of generating diverse, high-quality text outputs in creative writing by introducing Quality-Diversity through AI Feedback (QDAIF), which uses language models to guide an evolutionary algorithm for evaluation and variation, resulting in better coverage of the search space with high-quality samples compared to non-QD controls and showing reasonable agreement between AI and human evaluations.
In many text-generation problems, users may prefer not only a single response, but a diverse range of high-quality outputs from which to choose. Quality-diversity (QD) search algorithms aim at such outcomes, by continually improving and diversifying a population of candidates. However, the applicability of QD to qualitative domains, like creative writing, has been limited by the difficulty of algorithmically specifying measures of quality and diversity. Interestingly, recent developments in language models (LMs) have enabled guiding search through AI feedback, wherein LMs are prompted in natural language to evaluate qualitative aspects of text. Leveraging this development, we introduce Quality-Diversity through AI Feedback (QDAIF), wherein an evolutionary algorithm applies LMs to both generate variation and evaluate the quality and diversity of candidate text. When assessed on creative writing domains, QDAIF covers more of a specified search space with high-quality samples than do non-QD controls. Further, human evaluation of QDAIF-generated creative texts validates reasonable agreement between AI and human evaluation. Our results thus highlight the potential of AI feedback to guide open-ended search for creative and original solutions, providing a recipe that seemingly generalizes to many domains and modalities. In this way, QDAIF is a step towards AI systems that can independently search, diversify, evaluate, and improve, which are among the core skills underlying human society's capacity for innovation.