Chen Zhe Sage
Departments of Psychiatry, Neuroscience and Physiology, and Biomedical Engineering at the New York University, New York.
IEEE Signal Process Mag. 2024 Nov;41(6):94-104. doi: 10.1109/msp.2024.3484629. Epub 2025 Jan 1.
Rapid advances in generative artificial intelligence (AI) and deep representation learning have revolutionized numerous engineering applications in signal processing, computer vision, speech recognition and translation, and natural language processing due to amazingly powerful representation power (e.g., [1,2]). Generative AI-empowered tools, such as ChatGPT and Sora, have fundamentally changed the landscape of human-computer communications research. One emerging application along this line is to link the brain to the computer (i.e., brain-computer interface or BCI) and to develop paradigm-shift brain-to-content technologies. This BCI system upgrade (i.e., BCI 2.0) is empowered by generative AI and deep learning ("new engine") and large amounts of data ("gas"). In this article, we will revisit the old song sung in a new tune, highlight some state-of-the-art progresses, and briefly discuss the future outlook.
生成式人工智能（AI）和深度表征学习的快速发展，凭借其惊人强大的表征能力（例如，[1,2]），彻底改变了信号处理、计算机视觉、语音识别与翻译以及自然语言处理等众多工程应用。诸如ChatGPT和Sora等由生成式AI驱动的工具，从根本上改变了人机通信研究的格局。沿着这条线的一个新兴应用是将大脑与计算机连接起来（即脑机接口或BCI），并开发范式转变的脑到内容技术。这种BCI系统升级（即BCI 2.0）由生成式AI和深度学习（“新引擎”）以及大量数据（“燃料”）赋能。在本文中，我们将旧曲新唱，突出一些最新进展，并简要讨论未来展望。
