张雅楠zeyarna的博客

大模型人工智能 VS 神经技术人工智能

1. 大模型人工智能原理
   • 数学模型抽象后还原现实的过于简单，通过巨型参数进一步完善抽象数学模型
   • 诞生巨型软件公司
   • 因为海量参数导致的软件复杂性的高维护成本以及高运行算力要求
2. 神经技术人工智能原理
   • 基于人类经验而形成的人工智能，并且采用全仿真人类行为还原现实，及通过眼耳鼻口以及四肢与外界模糊交互及不断修正完成精确行为，改进人工智能程序不足，不断改进初始控制精确度
   • 诞生巨型学习创新知识公司
   • 对于人类知识抽取提纯后的简单知识库建立和后期的行为学习补偿程序，运算量极小，维护非常方便直接通过再次抽取和提纯人类新的学习结果更新知识库即可
3. 神经数据训练大模型：一种无奈的妥协
   • 因为现有成熟的商业产品的竞争获利要求
   • 神经数据训练大模型人工智能，一种比较廉价的高速市场化方案

Large model artificial intelligence VS Human neuro artificial intelligence （HNAI）

1. Principles of large model artificial intelligence
   • It is too simple to restore reality after abstraction of mathematical models, and further improve abstract mathematical models through giant parameters
   • Birth of giant software companies
   • High maintenance costs and high operating computing power requirements due to software complexity caused by massive parameters
2. Principles of Neurotechnology, Artificial Intelligence
   • Artificial intelligence based on human experience, and uses full simulation of human behavior to restore reality, and completes precise behavior through vague interaction with the outside world through eyes, ears, nose, mouth, and limbs, and continuous correction, improving the deficiencies of artificial intelligence programs and continuously improving the accuracy of initial control
   • Birth of a giant learning and innovation knowledge company
   • For the establishment of a simple knowledge base after human knowledge extraction and purification and the later behavioral learning compensation program, the amount of operation is very small, and the maintenance is very convenient, and the knowledge base can be updated directly by extracting and purifying new human learning results again
3. Neural data training large models: a helpless compromise
   • Because of the competitive profit requirements of existing mature commercial products
   • Neural data training large model artificial intelligence, a relatively cheap high-speed market-oriented solution