Mapping information and light: Trends of AI-enabled metaphotonics

Abstract

A dynamic convergence between metaphotonics and artificial intelligence (AI) is underway. In this review, AI is conceptualized as a tool for mapping input and output data. From this perspective, an analysis is conducted on how input and output data are set, aiming to discern the following three key trends in the utilization of AI within the field of metaphotonics. 1. The advancement of forward modeling and inverse design, utilizing AI for mapping metaphotonic device design and the corresponding optical properties. 2. Optical neural networks (ONNs), an emerging field that implements AI using metaphotonics by processing information within electromagnetic waves. 3. The field of metasensors, employing metamaterials to encode optical information for measurement and processing using AI to demonstrate high performance sensing. We round up the review with our perspectives on AI and metaphotonics research and discuss the future trends, challenges, and developments. © 2024 Elsevier Ltd