Alex Wissner-Gross: A new equation for intelligence

198,752 views ・ 2014-02-06

TED


请双击下面的英文字幕来播放视频。

翻译人员: Phillip Feng 校对人员: Zhiting Chen
00:12
Intelligence -- what is it?
0
12899
3667
智能--它是什么?
00:16
If we take a look back at the history
1
16566
2291
当我们回顾在历史上
00:18
of how intelligence has been viewed,
2
18857
2624
智能是如何被看待的,
00:21
one seminal example has been
3
21481
3618
一个开创性的例子是
00:25
Edsger Dijkstra's famous quote that
4
25099
3477
艾兹格•迪杰斯特拉的著名引述,
00:28
"the question of whether a machine can think
5
28576
3111
"关于一台机器能否思考的问题
00:31
is about as interesting
6
31687
1310
与关于
00:32
as the question of whether a submarine
7
32997
2971
一艘潜艇是否会游泳的问题
00:35
can swim."
8
35968
1790
几乎同样有趣"。
00:37
Now, Edsger Dijkstra, when he wrote this,
9
37758
3844
当艾兹格•迪杰斯特拉 写下这句话的时候,
00:41
intended it as a criticism
10
41602
2054
他的用意是去批判那些
00:43
of the early pioneers of computer science,
11
43656
3000
早年间开辟了计算机科学的先锋,
00:46
like Alan Turing.
12
46656
1747
比如阿兰 · 图灵。
00:48
However, if you take a look back
13
48403
2499
然而,如果你回顾过去
00:50
and think about what have been
14
50902
1965
并予以思考,有哪些
00:52
the most empowering innovations
15
52867
1996
最有利于发展的创新,
00:54
that enabled us to build
16
54863
1879
让我们有机会能够制造出
00:56
artificial machines that swim
17
56742
2234
会游泳的机器
00:58
and artificial machines that [fly],
18
58976
2573
和会[飞]的机器,
01:01
you find that it was only through understanding
19
61549
3547
你会发现,只有通过了解
01:05
the underlying physical mechanisms
20
65096
2608
游泳和飞行
01:07
of swimming and flight
21
67704
2779
背后的物理机制,
01:10
that we were able to build these machines.
22
70483
3172
我们才有能力去制造这些机器。
01:13
And so, several years ago,
23
73655
2256
所以说,在几年前,
01:15
I undertook a program to try to understand
24
75911
3249
我着手了一个项目, 试图去了解
01:19
the fundamental physical mechanisms
25
79160
2634
智能背后的
01:21
underlying intelligence.
26
81794
2768
基础物理机制。
01:24
Let's take a step back.
27
84562
1860
我们先退一步说。
01:26
Let's first begin with a thought experiment.
28
86422
3149
首先,让我们从一个思维实验开始。
01:29
Pretend that you're an alien race
29
89571
2854
假装你是一个外星人,
01:32
that doesn't know anything about Earth biology
30
92425
3041
你对地球上的生物学、
01:35
or Earth neuroscience or Earth intelligence,
31
95466
3116
神经科学和智能一无所知,
01:38
but you have amazing telescopes
32
98582
2192
但你有绝佳的望远镜,
01:40
and you're able to watch the Earth,
33
100774
2362
因此你能观望地球,
01:43
and you have amazingly long lives,
34
103136
2332
你的寿命也惊人地长,
01:45
so you're able to watch the Earth
35
105468
1499
所以你可以观察地球
01:46
over millions, even billions of years.
36
106967
3442
超过数百万年,甚至几十亿年。
01:50
And you observe a really strange effect.
37
110409
3015
然后你观察到一个很奇怪的现象。
01:53
You observe that, over the course of the millennia,
38
113424
4312
你观察到,几千年来,
01:57
Earth is continually bombarded with asteroids
39
117736
4285
地球不断地与小行星发生碰撞
02:02
up until a point,
40
122021
2087
直到某一刻,
02:04
and that at some point,
41
124108
1531
而在那一刻,
02:05
corresponding roughly to our year, 2000 AD,
42
125639
4192
大约对应的是公元2000年,
02:09
asteroids that are on
43
129831
1716
那些在地球撞击轨道
02:11
a collision course with the Earth
44
131547
1931
上的小行星,
02:13
that otherwise would have collided
45
133478
1975
本该相撞
02:15
mysteriously get deflected
46
135453
2415
但却被神秘地弹开了
02:17
or they detonate before they can hit the Earth.
47
137868
3072
或者在碰到地球之前就引爆了。
02:20
Now of course, as earthlings,
48
140940
2083
当然,作为地球人,
02:23
we know the reason would be
49
143023
1544
我们知道其中的原因是
02:24
that we're trying to save ourselves.
50
144567
1756
我们正试图自我拯救。
02:26
We're trying to prevent an impact.
51
146323
3080
我们要防止撞击发生。
02:29
But if you're an alien race
52
149403
1711
但如果你是一个外星人,
02:31
who doesn't know any of this,
53
151114
1146
对这些一无所知,
02:32
doesn't have any concept of Earth intelligence,
54
152260
2514
对地球上的智能也没有任何概念,
02:34
you'd be forced to put together
55
154774
1728
这就迫使你去总结
02:36
a physical theory that explains how,
56
156502
2918
一种物理理论, 去解释其原因,
02:39
up until a certain point in time,
57
159420
2538
直到在某一刻,
02:41
asteroids that would demolish the surface of a planet
58
161958
4449
本应摧毁一个星球表面的小行星,
02:46
mysteriously stop doing that.
59
166407
3231
神秘地停止了这种行为。
02:49
And so I claim that this is the same question
60
169638
4204
因此我声称这个问题
02:53
as understanding the physical nature of intelligence.
61
173842
3998
与理解智能的物理本质的问题 是相同的。
02:57
So in this program that I undertook several years ago,
62
177840
3882
因此,在我几年前着手的 这个项目中,
03:01
I looked at a variety of different threads
63
181722
2765
我研究了许多不同的线程,
03:04
across science, across a variety of disciplines,
64
184487
3162
跨越科学界,跨越多个学科,
03:07
that were pointing, I think,
65
187649
1892
在我看来,他们都指向
03:09
towards a single, underlying mechanism
66
189541
2548
一个统一的、潜在的
03:12
for intelligence.
67
192089
1581
智能机制。
03:13
In cosmology, for example,
68
193670
2546
例如在宇宙学中,
03:16
there have been a variety of different threads of evidence
69
196216
2747
就存在着各种各样的线索,
03:18
that our universe appears to be finely tuned
70
198963
3407
它们显示我们的宇宙就 为了智能的开发,
03:22
for the development of intelligence,
71
202370
2153
而被精准地调试过,
03:24
and, in particular, for the development
72
204523
2389
和特别是的对于发展
03:26
of universal states
73
206912
1886
世界各国
03:28
that maximize the diversity of possible futures.
74
208798
4098
去实现有最大多样化可能性的未来。
03:32
In game play, for example, in Go --
75
212896
2344
在棋牌界,举个例子,围棋--
03:35
everyone remembers in 1997
76
215240
3025
大家都记得在1997年的时候
03:38
when IBM's Deep Blue beat Garry Kasparov at chess --
77
218265
3951
IBM制作的机器人“深蓝“打败了 世界象棋冠军加里·卡斯帕罗夫--
03:42
fewer people are aware
78
222216
1523
很少有人意识到
03:43
that in the past 10 years or so,
79
223739
2018
在过去10年左右的时间里,
03:45
the game of Go,
80
225757
1198
围棋,
03:46
arguably a much more challenging game
81
226955
1956
可以说是一个更具挑战性的游戏,
03:48
because it has a much higher branching factor,
82
228911
2425
因为它具有更高的分支系数,
03:51
has also started to succumb
83
231336
1702
也已开始屈服于
03:53
to computer game players
84
233038
1865
电脑这个游戏对手,
03:54
for the same reason:
85
234903
1573
出于同样的原因:
03:56
the best techniques right now for computers playing Go
86
236476
2800
现在,电脑下围棋的 最佳技术方法
03:59
are techniques that try to maximize future options
87
239276
3696
是在下棋的过程中, 试图最大化
04:02
during game play.
88
242972
2014
未来的各种可能性。
04:04
Finally, in robotic motion planning,
89
244986
3581
最后,在机器人的运动规划中,
04:08
there have been a variety of recent techniques
90
248567
2182
有各种各样的新颖技术,
04:10
that have tried to take advantage
91
250749
1902
它们有试图利用
04:12
of abilities of robots to maximize
92
252651
3146
机器人的能力去将
04:15
future freedom of action
93
255797
1506
未来的行动自由最大化,
04:17
in order to accomplish complex tasks.
94
257303
3097
从而完成复杂的任务。
04:20
And so, taking all of these different threads
95
260400
2355
因此,考虑所有这些不同的线程
04:22
and putting them together,
96
262755
1622
并把它们放在一起,
04:24
I asked, starting several years ago,
97
264377
2640
从几年前开始我就在问,
04:27
is there an underlying mechanism for intelligence
98
267017
2850
有没有一种潜在的智能机制
04:29
that we can factor out
99
269867
1673
可以让我们分解出
04:31
of all of these different threads?
100
271540
1774
所有这些不同的线程?
04:33
Is there a single equation for intelligence?
101
273314
4593
是否存在一个 关于智能的公式?
04:37
And the answer, I believe, is yes. ["F = T ∇ Sτ"]
102
277907
3371
而我相信答案是有。 ["F = T ∇ SΤ"]
04:41
What you're seeing is probably
103
281278
1913
你看到的可能是
04:43
the closest equivalent to an E = mc²
104
283191
3294
我所见过最接近于 E = mc²
04:46
for intelligence that I've seen.
105
286485
2830
的智慧。
04:49
So what you're seeing here
106
289315
1702
所以你在这里看到的
04:51
is a statement of correspondence
107
291017
2669
是一张对应表,
04:53
that intelligence is a force, F,
108
293686
4435
其中智能是一种力量,F,
04:58
that acts so as to maximize future freedom of action.
109
298121
4650
它的作用是为了便于将未来的 行动自由最大化。
05:02
It acts to maximize future freedom of action,
110
302771
2375
它的作用是将未来的 行动自由最大化,
05:05
or keep options open,
111
305146
1628
或是保留灵活的选择权,
05:06
with some strength T,
112
306774
2225
与一种力量 T,
05:08
with the diversity of possible accessible futures, S,
113
308999
4777
和有多种可能性的、 可实现的未来,S,
05:13
up to some future time horizon, tau.
114
313776
2550
一直到某个未来的开始, tau(希腊字母)。
05:16
In short, intelligence doesn't like to get trapped.
115
316326
3209
简而言之,智能不喜欢被困住。
05:19
Intelligence tries to maximize future freedom of action
116
319535
3055
智能试图将未来的行动自由最大化,
05:22
and keep options open.
117
322590
2673
并保留选择权。
05:25
And so, given this one equation,
118
325263
2433
所以,鉴于这一公式,
05:27
it's natural to ask, so what can you do with this?
119
327696
2532
你自然会问, 那么这些可以让你做什么?
05:30
How predictive is it?
120
330228
1351
它是预测性有多高?
05:31
Does it predict human-level intelligence?
121
331579
2135
它能否预测人类的智能水平?
05:33
Does it predict artificial intelligence?
122
333714
2818
它能够预测人工智能吗?
05:36
So I'm going to show you now a video
123
336532
2042
因此,我将要展示给你们一段视频,
05:38
that will, I think, demonstrate
124
338574
3420
我认为,它会展示出
05:41
some of the amazing applications
125
341994
2288
单是这一个公式的
05:44
of just this single equation.
126
344282
2319
一些惊人的应用。
05:46
(Video) Narrator: Recent research in cosmology
127
346601
1979
(视频)讲述人: 宇宙学的最近研究
05:48
has suggested that universes that produce
128
348580
2047
反应了那些产生更多混乱、
05:50
more disorder, or "entropy," over their lifetimes
129
350627
3481
或者"熵"的宇宙, 在他们的生命中
05:54
should tend to have more favorable conditions
130
354108
2478
应该倾向于产生更多 有利的情况,
05:56
for the existence of intelligent beings such as ourselves.
131
356586
3016
让像我们这样的智慧生物 得以存在。
05:59
But what if that tentative cosmological connection
132
359602
2574
但假如那个在熵与智能之间
06:02
between entropy and intelligence
133
362176
1843
暂定的宇宙链接
06:04
hints at a deeper relationship?
134
364019
1771
暗示着更深层的关系呢?
06:05
What if intelligent behavior doesn't just correlate
135
365790
2564
如果智能的行为不仅只与
06:08
with the production of long-term entropy,
136
368354
1844
长期熵的生产相关,
06:10
but actually emerges directly from it?
137
370198
2318
而是直接由其产生的呢?
06:12
To find out, we developed a software engine
138
372516
2406
为了找到答案, 我们开发了一个软件引擎
06:14
called Entropica, designed to maximize
139
374922
2503
称为 Entropica, 设计的意图是将
06:17
the production of long-term entropy
140
377425
1768
长期熵的生产最大化,
06:19
of any system that it finds itself in.
141
379193
2576
无论它身在任何系统内。
06:21
Amazingly, Entropica was able to pass
142
381769
2155
惊人的是,Entropica 通过了
06:23
multiple animal intelligence tests, play human games,
143
383924
3456
多个动物的智能测验、 玩人类的游戏、
06:27
and even earn money trading stocks,
144
387380
2146
甚至在股票交易中赚钱,
06:29
all without being instructed to do so.
145
389526
2111
而且完全没有被给出那些指示。
06:31
Here are some examples of Entropica in action.
146
391637
2518
下面是一些 Entropica 的行动实例。
06:34
Just like a human standing upright without falling over,
147
394155
3205
就像人类站立不会跌到,
06:37
here we see Entropica
148
397360
1230
这里我们可以看到 Entropica
06:38
automatically balancing a pole using a cart.
149
398590
2885
自动地使用购物车去平衡棍子。
06:41
This behavior is remarkable in part
150
401475
2012
这种行为可以说是非常卓越的
06:43
because we never gave Entropica a goal.
151
403487
2331
因为我们从未给 Entropica 设定一个目标。
06:45
It simply decided on its own to balance the pole.
152
405818
3157
它自己就决定去平衡那根棍子。
06:48
This balancing ability will have appliactions
153
408975
2132
这种平衡能力将能应用于
06:51
for humanoid robotics
154
411107
1397
人形机器人
06:52
and human assistive technologies.
155
412504
2515
和人类的辅助科技。
06:55
Just as some animals can use objects
156
415019
2001
正如一些动物可以使用
06:57
in their environments as tools
157
417020
1442
环境中的物体作为工具
06:58
to reach into narrow spaces,
158
418462
1987
去伸入狭窄的空间,
07:00
here we see that Entropica,
159
420449
1882
这里我们可以看到 Entropica,
07:02
again on its own initiative,
160
422331
1838
同样是自主的,
07:04
was able to move a large disk representing an animal
161
424169
2910
能够移动一个表示动物的大圆盘
07:07
around so as to cause a small disk,
162
427079
2345
去把一个表示工具的小圆盘,
07:09
representing a tool, to reach into a confined space
163
429424
2771
去深入一个狭窄的空间,
07:12
holding a third disk
164
432195
1537
那里有第三个圆盘,
07:13
and release the third disk from its initially fixed position.
165
433732
2972
并把第三个圆盘从它初始 的静态解放出来.
07:16
This tool use ability will have applications
166
436704
2189
这种工具的使用能力将能运用于
07:18
for smart manufacturing and agriculture.
167
438893
2359
智能制造业和农业。
07:21
In addition, just as some other animals
168
441252
1944
此外,正如其他一些动物
07:23
are able to cooperate by pulling opposite ends of a rope
169
443196
2696
能够合作起来同时去拉 一根绳子的两端
07:25
at the same time to release food,
170
445892
2053
从而释放食物,
07:27
here we see that Entropica is able to accomplish
171
447945
2295
这里我们可以看到 Entropica 有能力完成
07:30
a model version of that task.
172
450240
1988
这项任务的模型版本。
07:32
This cooperative ability has interesting implications
173
452228
2522
这种合作能力能够带来有趣的影响,
07:34
for economic planning and a variety of other fields.
174
454750
3435
在经济规划和各种其他领域中。
07:38
Entropica is broadly applicable
175
458185
2071
Entropica 可以广泛适用于
07:40
to a variety of domains.
176
460256
1943
各种各样的领域。
07:42
For example, here we see it successfully
177
462199
2442
例如,在这里我们看到它成功的
07:44
playing a game of pong against itself,
178
464641
2559
与自己玩乒乓球游戏,
07:47
illustrating its potential for gaming.
179
467200
2343
说明其在游戏界的潜力。
07:49
Here we see Entropica orchestrating
180
469543
1919
在这里我们看到 Entropica 指挥着
07:51
new connections on a social network
181
471462
1839
社交网络上新的关系,
07:53
where friends are constantly falling out of touch
182
473301
2760
在这朋友们不断的失去联系
07:56
and successfully keeping the network well connected.
183
476061
2856
并成功地保持有效的网络连接。
07:58
This same network orchestration ability
184
478917
2298
这种相同的网络指挥能力
08:01
also has applications in health care,
185
481215
2328
在医疗保健、能源、和智能方面
08:03
energy, and intelligence.
186
483543
3232
都有相关的应用。
08:06
Here we see Entropica directing the paths
187
486775
2085
这里我们可以看到 Entropica
08:08
of a fleet of ships,
188
488860
1486
指挥一支舰队的路径,
08:10
successfully discovering and utilizing the Panama Canal
189
490346
3175
成功地发现并利用巴拿马运河,
08:13
to globally extend its reach from the Atlantic
190
493521
2458
然后将其范围从大西洋到太平洋
08:15
to the Pacific.
191
495979
1529
全球性地扩大。
08:17
By the same token, Entropica
192
497508
1727
同样的,Entropica
08:19
is broadly applicable to problems
193
499235
1620
可以广泛地适用于
08:20
in autonomous defense, logistics and transportation.
194
500855
5302
自主防御、 物流和运输地应用。
08:26
Finally, here we see Entropica
195
506173
2030
最后,在这里我们看到 Entropica
08:28
spontaneously discovering and executing
196
508203
2723
自主地发现和执行
08:30
a buy-low, sell-high strategy
197
510926
2067
一个低买高卖的策略,
08:32
on a simulated range traded stock,
198
512993
2178
这是在模拟的范围交易股票上,
08:35
successfully growing assets under management
199
515171
2331
它成功地将其管理的资产
08:37
exponentially.
200
517502
1424
成指数升涨。
08:38
This risk management ability
201
518926
1308
这种风险管理的能力
08:40
will have broad applications in finance
202
520234
2487
将在金融和保险领域
08:42
and insurance.
203
522721
3328
有广泛的应用。
08:46
Alex Wissner-Gross: So what you've just seen
204
526049
2091
阿历克斯•维斯纳-格罗斯: 你刚看到的
08:48
is that a variety of signature human intelligent
205
528140
4392
是各种具有代表性的人类智能
08:52
cognitive behaviors
206
532532
1757
的认知行为,
08:54
such as tool use and walking upright
207
534289
2831
例如工具的使用、直立行走
08:57
and social cooperation
208
537120
2029
和社会合作,
08:59
all follow from a single equation,
209
539149
2972
它们都遵循一个公式,
09:02
which drives a system
210
542121
1932
该公式所驱动的系统
09:04
to maximize its future freedom of action.
211
544053
3911
是要将其未来的行动自由最大化。
09:07
Now, there's a profound irony here.
212
547964
3007
现在,这里存在一个深刻的讽刺。
09:10
Going back to the beginning
213
550971
2024
回到最初,
09:12
of the usage of the term robot,
214
552995
3273
机器人这个术语的用法,
09:16
the play "RUR,"
215
556268
2903
"RUR,"这出戏,
09:19
there was always a concept
216
559171
2235
总存在一种概念就是
09:21
that if we developed machine intelligence,
217
561406
3226
如果我们开发了机器智能
09:24
there would be a cybernetic revolt.
218
564632
3027
就会产生一个人工智能的叛变。
09:27
The machines would rise up against us.
219
567659
3551
机器会奋起反抗我们。
09:31
One major consequence of this work
220
571210
2319
这项工作的主要成果之一
09:33
is that maybe all of these decades,
221
573529
2769
就是也许这几十年间,
09:36
we've had the whole concept of cybernetic revolt
222
576298
2976
我们对于人工智能的叛变 的整个概念
09:39
in reverse.
223
579274
2011
是颠倒的。
09:41
It's not that machines first become intelligent
224
581285
3279
机器不是先有了智慧
09:44
and then megalomaniacal
225
584564
2015
然后才变得狂妄
09:46
and try to take over the world.
226
586579
2224
并试图接管世界的。
09:48
It's quite the opposite,
227
588803
1434
其实几乎是相反的,
09:50
that the urge to take control
228
590237
2906
那种迫切的欲望,
09:53
of all possible futures
229
593143
2261
想要控制所有未来的所有可能
09:55
is a more fundamental principle
230
595404
2118
是比智能更基本的
09:57
than that of intelligence,
231
597522
1363
一个原则,
09:58
that general intelligence may in fact emerge
232
598885
3700
综合智能事实上可能是从
10:02
directly from this sort of control-grabbing,
233
602585
3559
这种控制欲中直接产生的,
10:06
rather than vice versa.
234
606144
4185
而不是反之。
10:10
Another important consequence is goal seeking.
235
610329
3769
另一个重要的成果是寻找目标。
10:14
I'm often asked, how does the ability to seek goals
236
614098
4360
我经常被问道, 寻找目标的能力
10:18
follow from this sort of framework?
237
618458
1620
怎么会遵循这种框架结构呢?
10:20
And the answer is, the ability to seek goals
238
620078
3028
答案是,寻找目标的能力
10:23
will follow directly from this
239
623106
1882
将直接遵循它,
10:24
in the following sense:
240
624988
1834
道理是这样的:
10:26
just like you would travel through a tunnel,
241
626822
2865
就像你要穿过一条隧道,
10:29
a bottleneck in your future path space,
242
629687
2505
你未来道路空间中的一个瓶颈,
10:32
in order to achieve many other
243
632192
1871
为了在以后实现许多
10:34
diverse objectives later on,
244
634063
2021
其他的各种目标,
10:36
or just like you would invest
245
636084
2372
或者就像你会投资
10:38
in a financial security,
246
638456
1787
于金融证券,
10:40
reducing your short-term liquidity
247
640243
2237
减少你的短期流动性
10:42
in order to increase your wealth over the long term,
248
642480
2400
从而长远的增加你的财富,
10:44
goal seeking emerges directly
249
644880
2337
目标的寻求直接涌现于
10:47
from a long-term drive
250
647217
1729
长期的驱动,
10:48
to increase future freedom of action.
251
648946
4037
为了增加未来的行动自由。
10:52
Finally, Richard Feynman, famous physicist,
252
652983
3528
最后,理查德 · 费曼, 这位著名的物理学家,
10:56
once wrote that if human civilization were destroyed
253
656511
3672
曾经写道, 如果人类文明被摧毁
11:00
and you could pass only a single concept
254
660183
1893
并且你只能将一个概念
11:02
on to our descendants
255
662076
1371
传承给我们的后代,
11:03
to help them rebuild civilization,
256
663447
2307
来帮助他们重建文明,
11:05
that concept should be
257
665754
1686
这个概念应该是
11:07
that all matter around us
258
667440
1852
我们身边的一切物质
11:09
is made out of tiny elements
259
669292
2323
都是由微小的元素组成的,
11:11
that attract each other when they're far apart
260
671615
2508
它们之间距离远的时候 会相互吸引,
11:14
but repel each other when they're close together.
261
674123
3330
但在靠的很近时 它们会互相排斥。
11:17
My equivalent of that statement
262
677453
1781
我与这句话等同的声明,
11:19
to pass on to descendants
263
679234
1268
来传递给后代,
11:20
to help them build artificial intelligences
264
680502
2712
帮助他们建立人工智能
11:23
or to help them understand human intelligence,
265
683214
2949
或是帮助他们理解 人类的智慧,
11:26
is the following:
266
686163
1267
是如下的话:
11:27
Intelligence should be viewed
267
687430
2053
智能应该被看作是
11:29
as a physical process
268
689483
1413
一个物理过程,
11:30
that tries to maximize future freedom of action
269
690896
2965
它试图将未来的行动自由最大化
11:33
and avoid constraints in its own future.
270
693861
3616
并且避免在自己的未来中的约束。
11:37
Thank you very much.
271
697477
1358
非常感谢。
11:38
(Applause)
272
698835
4000
(掌声)
关于本网站

这个网站将向你介绍对学习英语有用的YouTube视频。你将看到来自世界各地的一流教师教授的英语课程。双击每个视频页面上显示的英文字幕,即可从那里播放视频。字幕会随着视频的播放而同步滚动。如果你有任何意见或要求,请使用此联系表与我们联系。

https://forms.gle/WvT1wiN1qDtmnspy7