Generalist Multimodal AI: A Review of Architectures, Challenges and Opportunities

TL;DR

This paper reviews the development of generalist multimodal models, analyzing architectures, challenges, and opportunities to guide future research in creating versatile AI systems across multiple data modalities.

Contribution

It introduces a novel taxonomy based on architecture and training configurations, focusing on Unifiability, Modularity, and Adaptability for GMMs.

Findings

Identifies key architectural factors for GMMs

Highlights challenges in extending models beyond text and vision

Provides a roadmap for future multimodal AI research

Abstract

Multimodal models are expected to be a critical component to future advances in artificial intelligence. This field is starting to grow rapidly with a surge of new design elements motivated by the success of foundation models in natural language processing (NLP) and vision. It is widely hoped that further extending the foundation models to multiple modalities (e.g., text, image, video, sensor, time series, graph, etc.) will ultimately lead to generalist multimodal models, i.e. one model across different data modalities and tasks. However, there is little research that systematically analyzes recent multimodal models (particularly the ones that work beyond text and vision) with respect to the underling architecture proposed. Therefore, this work provides a fresh perspective on generalist multimodal models (GMMs) via a novel architecture and training configuration specific taxonomy. This includes factors such as Unifiability, Modularity, and Adaptability that are pertinent and essential to the wide adoption and application of GMMs. The review further highlights key challenges and prospects for the field and guide the researchers into the new advancements.