The rapid evolution of artificial intelligence presents not only unprecedented opportunities but also significant technical challenges, particularly in the development of next-generation computing hardware. To overcome these hurdles, there is an urgent demand for novel chip architectures that offer both ultralow power consumption and high computational efficiency. Neuromorphic computing, inspired by the neural architecture of the human brain, represents a paradigm shift beyond the conventional von Neumann framework, promising remarkable gains in processing capability. Here, we report an ambipolar transistor array based on a vertically stacked polymer/oxide heterostructure, meticulously engineered to integrate electrical computation with optical sensing within a single device. This transistor enables simultaneous electrical and optical modulation, supporting both synaptic transmission under electrical stimuli and dynamic visual information processing under optical inputs. The integrated array system demonstrates efficient and low-power execution of visual processing, classification, and prediction tasks, highlighting its potential for neuromorphic computing applications such as real-time traffic analysis. Our findings pave the way for multifunctional and energy-efficient neuromorphic hardware capable of bridging the gap between sensing and computation.
Zhong et al. (Fri,) studied this question.