当前位置:网站首页>Discussion | has large AI become autonomous? Lecun, chief scientist of openai
Discussion | has large AI become autonomous? Lecun, chief scientist of openai
2022-07-04 10:46:00 【Zhiyuan community】
lately ,OpenAI Chief scientist A word of , The whole machine learning circle is fried .
2 month 10 Japan , Elijah · Sartsqueville (Ilya Sutskever) A tweet means :「 Now the large neural network may have a weak sense of autonomy .」

The argument lasted for a few days .
2 month 13 Japan ,LeCun Even end up in person , Extremely firm and simple :
「 incorrect .」
「 even to the extent that 『 Weak consciousness 』 In the table, trace adjectives , and 『 Large neural networks 』 The adjective in represents large , All wrong . I think this needs a large-scale architecture that all neural networks do not have at present .」

OpenAI Chief scientist : Shows curiosity and openness to mysterious ideas , use 「 Probably 」 Start the sentence ; Meta chief AI scientists : The opening sentence is arbitrary 「 incorrect 」. Maybe this can explain the past 5 Mid year performance . dear Meta AI Researchers , This is my email address , Our company is hiring now .

A company can indeed build faster planes every day 、 Break the flight altitude record , Doing these things can make your name often make headlines , These breakthroughs may also be of some real use . But if the goal is to really break through the atmosphere and fly into space orbit , Then it needs to be in the condensate storage tank 、 Turbopump has worked hard on these unremarkable technologies . If you don't believe it, ask musk .

AI Researchers and students in the world , If you don't know yet , The following is the outline of this event : Except for the old-fashioned AI Companies spray each other and fight with each other , There is no meaningful substance in this matter . You'd better ask quality questions 、 Be curious 、 At the same time, maintain a suspicious attitude .
边栏推荐
- [machine] [server] Taishan 200
- Safety reinforcement learning based on linear function approximation safe RL with linear function approximation translation 1
- Dos:disk operating system, including core startup program and command program
- Latex learning insertion number - list of filled dots, bars, numbers
- From programmers to large-scale distributed architects, where are you (I)
- Virtual machine configuration network
- Network connection (II) three handshakes, four waves, socket essence, packaging of network packets, TCP header, IP header, ACK confirmation, sliding window, results of network packets, working mode of
- [Galaxy Kirin V10] [server] failed to start the network
- Hidden C2 tunnel -- use of icmpsh of ICMP
- Common system modules and file operations
猜你喜欢

C language - stack

Learning XML DOM -- a typical model for parsing XML documents

Huge number multiplication (C language)

【Day2】 convolutional-neural-networks
![[Galaxy Kirin V10] [desktop] build NFS to realize disk sharing](/img/72/5e725a44a50f152b477a4b2907a2d0.jpg)
[Galaxy Kirin V10] [desktop] build NFS to realize disk sharing

Error C4996 ‘WSAAsyncSelect‘: Use WSAEventSelect() instead or define _ WINSOCK_ DEPRECATED_ NO_ WARN

Introduction to tree and binary tree

What is an excellent architect in my heart?

Safety reinforcement learning based on linear function approximation safe RL with linear function approximation translation 1

Article publishing experiment
随机推荐
/*The rewriter outputs the contents of the IA array. It is required that the type defined by typedef cannot be used in the outer loop*/
Time complexity and space complexity
If you don't know these four caching modes, dare you say you understand caching?
MPLS: multi protocol label switching
Sword finger offer 05 (implemented in C language)
Vs201 solution to failure to open source file HPP (or link library file)
Realsense d435 d435i d415 depth camera obtains RGB map, left and right infrared camera map, depth map and IMU data under ROS
Hidden C2 tunnel -- use of icmpsh of ICMP
When I forget how to write SQL, I
DML statement of MySQL Foundation
Summary of several job scheduling problems
Application and Optimization Practice of redis in vivo push platform
Introduction to tree and binary tree
Error C4996 ‘WSAAsyncSelect‘: Use WSAEventSelect() instead or define _ WINSOCK_ DEPRECATED_ NO_ WARN
Knapsack problem and 0-1 knapsack problem
[Galaxy Kirin V10] [desktop] FTP common scene setup
OSPF comprehensive experiment
Safety reinforcement learning based on linear function approximation safe RL with linear function approximation translation 2
Sword finger offer 31 Stack push in and pop-up sequence
Doris / Clickhouse / Hudi, a phased summary in June
