Skip to content

Latest commit

 

History

History
78 lines (51 loc) · 4.86 KB

hybrid_AI.md

File metadata and controls

78 lines (51 loc) · 4.86 KB
Raw

AI must use hybrid technology (Take Tesla Autopilot as an example)

Copyright © 2021.10.12 Lin Pengcheng. All rights reserved.

I believe It is wrong that Tesla’s autopilot technology insists on using only visual technology and not using radar and other technologies at the same time.

They are the most important way to improve AI:

  1. the mutual verification of different information sources. Human intelligence is to integrate multiple sources of information to make decisions. This is common sense.
  2. the discovery and processing of contradictory information.
  3. Each technology has its advantages and limits. When it is developed to a certain level, the marginal cost of continuing to improve the level of a single technology is very high, and the complementary advantages of multiple technologies are cost-effective.

Tesla only uses visual technology. Although it is currently leading, but I believe that this is a narrow path. In order to save a little cost at the moment, it is taking the wrong path in the general direction.

My explainable AI framework based on legal models and warehouse/workshop models based on management principles is best at organizing and coordinating multiple hybrid technologies to solve problems in the best way. Because the legal sciences are good at solving differences and contradictions, the management sciences are good at organizing and coordinating, complementing each other's strengths, and cooperating to complete tasks. It is similar to the heterogeneous parallel computing of the computer, and this is the gifted magical power of the warehouse/workshop model.

Postscript:

After Apple, Intel, Microsoft, ARM and others, Tesla is following my theory.

Reference:

AI也要兼听则明（以特斯拉自动驾驶为例）

版权所有 © 2021.10.12 林鹏程， 保留所有权利。

我认为特斯拉的自动驾驶技术坚持只用视觉技术,没有同时使用雷达等技术的方法是错误的.

改善AI的最重要的途径:

  1. 不同信息来源的互参互证, 人类智能就是融合多种信息来源进行决策, 这是常识.
  2. 矛盾信息的发现和处理.
  3. 每种技术都有其优势和极限, 发展到一定水平, 继续提高单项技术水平的边际成本很高, 多种技术优势互补才符合成本效益原则.

特斯拉只使用视觉技术, 虽然目前暂时领先, 可偏听偏信这是路子是走窄了, 为了当下节约一点成本，在大方向上走错路了.

我的基于法律模型和基于管理原则的仓库/车间模型的可解释AI框架， 最善长组织和协调多种混合技术以最优方法解决问题。 因为法学善长解决分岐和矛盾, 管理科学最善长组织协调, 优势互补, 合作完成任务。 它类似于计算机的异构并行计算, 它是仓库/车间模型的天赋神通.

后记:

言出法随，继＂苹果，Intel, 微软，ARM＂等公司之后，＂特斯拉＂追随我的理论.

2022.12.13, InfoQ新闻, 《放出狠话又自打脸，特斯拉将重新在车辆上安装雷达，最快下个月发布》.

参考文献: