Full-analogue Photonic AI for Embracing the Uncertainty of the Environment
Alеxander Raikov, Meng Guo · 2025
The chapter aims to accelerate understanding of the environment’s uncertainty by creating a full-analogue photonic artificial intelligence (PAI) alternative to digital artificial intelligence (AI). PAI design draws on the study of light’s role in the evolution of life on Earth. It considers approaches such as inverse problem-solving, optical Fourier convolutions, thermodynamics, and synthesising photonic materials for rewritable 3D holographic memory using protein. PAI preserves the natural signal’s spectrum, providing a more complete reflection of hybrid (human–machine) reality. The PAI allows replacing the traditional multistep machine learning (ML) training of a multilayer neural network with a single-step simultaneous optical Fourier convolution of a set of training images and recording the result in separate dots (cells) of the 3D holographic matrix, which can store many allowed quantum states in a single cell. PAI provides advanced possibilities to state authorities, researchers, professionals, individuals, and society. PAI can speed up digital ML by several orders of magnitude, increasing the accuracy of understanding the environment’s uncertainty. The main research limitations include developing a reliable photonic material for the 3D holographic memory, optical deflectors, and analogue–digital interface.