与儿子聊天记录

我也有机会跟儿子网聊了，用iMessage，哼。

母：       "Did you read the article I sent to you, about math is not perfect?"

子：       It's interesting,

but I don’t think it meant math isn’t perfect

it meant that math doesn’t represent the world

but that’s not necessarily because math isn’t perfect, but the world isn’t (mathematically) perfect

母：       "Ricardo said these days, students are learning less about using accurate math to describe the world"

"People use computers to simulate the world "

子：       well classical mechanics is basically a set of formulas that are all solved
except for three-body problems

no, as in technically people don’t need to know it if it’s in a physics book, people can always look it up when they need it

母：       "I see what you mean. Basically, new generation could pay attention to different things ... What do you think students should learn now?"

子：       I think the problem with math education that Ricardo was talking about is more a problem about the size of mathematics than the interests of students

because modern math research is so complicated that it requires many years of study

like maybe people have to develop AI or augment human brains in order to speed up comprehension of math

so computer science is important, as is neuroscience/psychology, formal logic

of course in order to run society we still need politicians, economists, doctors etc

母：       "I think another problem is that mathematicians work too hard in math, they don't have chance to connect their knowledge to applications. It turned out, most of the science and engineering work is still using 19th century math"

子：       yeah there’s an overspecialization of scientists now because there’s so much knowledge in each little subject

but something like Watson could be good to make the connections

 

比较核心的一段话，我当时没理解，现在回头看很有意思。就是说，“人们发展人工智能，或者说提高人脑的能力，以此来加速数学学习。”这是新一代孩子的视角。我们学习的时候碰到困难，会觉得应该更加努力，头悬梁锥刺骨，或者觉得自己智慧不够，懊恼先天不足。新孩子的想法是，我们人类的大脑就是能力不足，我们需要人工智能的帮助。
 

所有跟帖： 

• 看了这样深度对话感到惭愧呀 -simplylove- ♀ (199 bytes) () 01/27/2016 postreply 15:04:56

• 你说的这种对话在我这儿也占大多数，就不记录了。为了让对方多回答几个字，我折磨自己想各种法子 -hobo-hobo- ♀ (61 bytes) () 01/27/2016 postreply 15:08:27

• 改变社会的两大途径，技术和政治。 -大号蚂蚁- ♂ (0 bytes) () 01/27/2016 postreply 15:17:09

• 技术涵盖范围就大了。他们讨论的是数学的发展和作用。 -simplylove- ♀ (0 bytes) () 01/27/2016 postreply 15:23:59

• 这两天发现，我们攻关项目的复杂程度大概已经超出几个工作人员大脑的思考能力，AI才是出路 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 15:39:13

• 回头看孩子的思路，早就朝那个方向想了。才知道很多时候跟孩子聊天儿，自己都没聊明白 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 15:41:09

• 但是 ai 还是人设计出来的，不会超过人脑智力的。只能起辅助作用。 -simplylove- ♀ (0 bytes) () 01/27/2016 postreply 15:41:31

• 我以前也是这么想的，昨天开始改变了。人脑是有限，现在机器识别人脸，识别手写体的能力都已经超过了人。 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 15:49:06

• 就是说在经过足够训练之后，learning system可以做很多人做不了的事。目前系统还是人设计的，以后连系统都是系统设计的了 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 15:53:20

• 俺的理解是 -msrbcm- ♀ (109 bytes) () 01/27/2016 postreply 15:52:05

• 还不是这种，更像是一个模型在足够复杂之后，能表达或者构造出比人脑表达或构造的更复杂的体系 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 15:56:23

• 如此说来，以后真的会象MATRIX或者终结者一样，被电脑统治吗？ -msrbcm- ♀ (0 bytes) () 01/27/2016 postreply 15:58:37

• 我觉得统治应该还不至于。机器能做更复杂的东西在于专一，它就做那一件或几件事。 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 16:09:16

• 如此俺放心不少。：））） -msrbcm- ♀ (0 bytes) () 01/27/2016 postreply 16:23:36

• 经典几何可以用计算机证明。FFT，有限元，样条函数使计算机有效解决工程问题 -mondayadvisor- ♀ (0 bytes) () 01/27/2016 postreply 16:02:22

• 计算机辅助存在已久，以前可能比较单一，现在构造的新系统往往能给出惊奇的结果 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 16:12:16

• 构造复杂新系统不能于创新。learning, recognition 还是基于人设定的 rules. 不了解其它行业情况，机器人 -simplylove- ♀ (53 bytes) () 01/27/2016 postreply 16:21:01

• 据说分析太空星球数据，可以提出宇宙起源的新模型。当然路还长。 -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 16:36:32

• 母子都棒棒哒：） -princessonthepea- ♀ (0 bytes) () 01/27/2016 postreply 16:46:26

• 顿时回到一年级 :-) -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 16:55:00

• 我也就这水平，呵呵：-） -princessonthepea- ♀ (0 bytes) () 01/27/2016 postreply 17:22:58

• 与水平无关，语气很嗲 :-) -hobo-hobo- ♀ (0 bytes) () 01/27/2016 postreply 17:26:00