Can we learn to talk to sperm whales? | David Gruber | TED

73,713 views ・ 2021-04-28

TED


请双击下面的英文字幕来播放视频。

翻译人员: Qingyue Sun 校对人员: Helen Chang
00:12
You are about to hear the sounds of the largest-toothed predator
0
12375
3309
你即将听到
地球上最大的有齿肉食动物的声音。
00:15
on the planet:
1
15708
1268
这种比校车还大的动物,
00:17
an animal bigger than a school bus
2
17000
2351
00:19
with perhaps the most sophisticated form of communication
3
19375
3059
也许拥有现有的
00:22
that has ever existed.
4
22458
2310
最复杂的交流方式。
00:24
(Video: whale clicking)
5
24792
6000
(视频:鲸鱼发出的滴答声)
00:43
These are the sounds of the mighty sperm whale,
6
43833
2518
这是健壮的抹香鲸的声音。
00:46
a fellow mammal that can dive almost a mile,
7
46375
2893
它是一种能下潜 将近一英里的哺乳动物,
00:49
hold its breath for more than an hour
8
49292
2267
能屏住呼吸超过一小时,
00:51
and lives in these amazingly complex, matriarchal societies.
9
51583
3851
并且住在奇迹般复杂的母系社会里。
00:55
These clicks you heard,
10
55458
1310
你所听到的这些滴答声
00:56
called codas,
11
56792
1267
被称作尾波,
00:58
are just a facet of what we know of their communication.
12
58083
3101
只是我们所知的 它们交流方式的一部分。
01:01
We know these animals are communicating,
13
61208
2393
我们知道这些动物在交流,
01:03
we just don't yet know what they're saying.
14
63625
2976
但只是还不知道它们在说什么。
01:06
Project CETI aims to find out.
15
66625
2268
CETI项目希望发现这个奥秘。
01:08
Over the next five years,
16
68917
1559
在未来的五年里,
01:10
our team of AI specialists,
17
70500
1684
我们团队的AI专家们,
01:12
roboticists, linguists
18
72208
1768
机器专家们,语言学家们,
和海洋生物学家们,
01:14
and marine biologists
19
74000
1434
01:15
aim to use the most cutting-edge technologies
20
75458
2310
希望用最先进的技术
01:17
to make contact with another species,
21
77792
1892
来和其他物交流,
01:19
and hopefully communicate back.
22
79708
3476
希望得到回应。
01:23
We believe that by listening deeply to nature,
23
83208
2518
我们相信,通过用心地聆听自然,
01:25
we can change our perspective of ourselves
24
85750
2476
我们能改变我们对自身的看法,
01:28
and reshape our relationship with all life on this planet.
25
88250
5059
并重塑我们与地球上其他生物的关系。
01:33
This of course seems like an impossible goal.
26
93333
3393
这当然听起来是个不可能实现的目标。
01:36
People have been trying to make contact with other animals
27
96750
2768
数百年里,人们尝试着 和其他动物沟通。
01:39
for hundreds of years.
28
99542
1309
01:40
How could we do what others could not,
29
100875
2393
我们如何做到别人做不到的,
01:43
especially given that I'm sitting here on my couch in New York City
30
103292
4142
特别是当我现在坐在 纽约市里我的沙发上,
01:47
in the middle of a pandemic and protests?
31
107458
2476
在疫情和抗议之中?
01:49
I've spent the last 20 years as a marine biologist and oceanographer,
32
109958
3810
二十年来我一直是个 海洋生物学家和海洋摄影师,
01:53
studying the ocean from all different perspectives,
33
113792
3059
从不同角度学习着海洋,
01:56
from microbes to sharks.
34
116875
1851
包括从微生物到鲨鱼的角度。
01:58
I've assembled interdisciplinary teams
35
118750
1893
我组织过跨学科的团队,
02:00
that have built the first shark-eye camera
36
120667
2142
建造了第一个鲨鱼眼摄像机
02:02
to see the world from a shark's perspective,
37
122833
2226
来通过鲨鱼的视角看世界,
02:05
and have collaborated with engineers
38
125083
1851
并且和工程师们合作过
02:06
to design robots so gentle that they don't even stress a jellyfish.
39
126958
4893
来设计轻柔到不会让水母紧张的机器。
02:11
But it wasn't until 2018
40
131875
2143
但直到2018年,当我在
02:14
when I was on fellowship
41
134042
1309
拉德克利夫高等研究所做研究员时,
02:15
at the Radcliffe Institute for Advanced Study
42
135375
2434
02:17
that I realized that perhaps the best way to understand the ocean
43
137833
3060
我才意识到也许理解海洋
02:20
and its inhabitants
44
140917
1267
和它的居民的最好方式
02:22
wasn't just by seeing the world through their eyes,
45
142208
3101
不只是通过它们的视角看世界,
02:25
but by listening --
46
145333
1310
而是通过倾听——
02:26
by really, deeply listening.
47
146667
2142
通过真正、用心的聆听。
02:28
I became interested in sperm whales when I heard their sounds.
48
148833
3060
当我听到抹香鲸的声音时 我开始对它们感兴趣。
02:31
They sounded like they were coming from another universe;
49
151917
2726
它们听起来像来自另一个宇宙,
02:34
a siren song being broadcast from the darkest reaches of the sea.
50
154667
4434
是海洋至黑处传来的海妖的歌声。
02:39
These weren't the typical harmonious whale songs
51
159125
3393
这些不是我所习惯的,
02:42
that I had been accustomed to.
52
162542
1559
鲸鱼典型的和谐歌声。
02:44
These sounded more like digital data transfer.
53
164125
3268
这些听起来更像电子数据传播声。
02:47
We assembled the future Project CETI team
54
167417
2434
我们组建了现在的CETI项目团队,
02:49
and began discussing how to use the most advanced technologies
55
169875
3351
并开始讨论如何使用最先进的技术
02:53
to communicate with whales.
56
173250
2101
来与鲸鱼沟通。
02:55
One of the principal conclusions
57
175375
1559
主要结论之一是
02:56
was that machine learning had a really good chance
58
176958
2685
机器学习有很好的机会
02:59
of understanding the patterns of sperm whale communication.
59
179667
2934
去理解抹香鲸交流的规律。
03:02
And the time to apply these technologies was now.
60
182625
3684
现在是应用这些技术的时候了。
03:06
Cracking the interspecies communication code
61
186333
2726
破解物种间的通讯密码
03:09
didn't just seem possible,
62
189083
3018
不仅看起来有可能实现,
03:12
it almost seemed inevitable.
63
192125
2268
并且几乎是不可避免的。
03:14
But how can analyzing patterns help us converse with whales
64
194417
2851
但是分析规律如何帮助我们与鲸鱼
03:17
and other animals?
65
197292
1267
和其他动物沟通呢?
03:18
Well, step one is to understand the elements of sperm whale communication.
66
198583
4393
第一步是了解抹香鲸交流的要素。
你们听到的这些尾波 不是我们所认为的句子,
03:23
These codas you heard don't appear to be sentences as we know them,
67
203000
4059
03:27
but there's clear structure in how these animals communicate.
68
207083
2935
但这些动物的交流方式有明确的结构。
03:30
Sperm whales send codas back and forth to each other
69
210042
2934
抹香鲸有序地互相发送
尾波,
03:33
in sequences,
70
213000
1309
03:34
and there are regional dialects like British and Australian accents.
71
214333
3976
还有地域方言, 就像英国和澳大利亚的口音。
03:38
This is exactly why machine learning is such a powerful tool.
72
218333
3768
这就是为什么机器学习 是如此强大的工具。
03:42
These approaches analyze patterns in relationship and map meaning to them.
73
222125
4018
这些方法分析规律间的关联 并得出意义。
03:46
Just a few years ago, scientists used machine learning
74
226167
2767
就在几年前,科学家们通过机器学习
03:48
to translate between two totally unknown human languages.
75
228958
3810
翻译了两种完全未知的人类语言。
03:52
Not by using a Rosetta Stone or a dictionary,
76
232792
2767
不是用罗塞塔石碑或字典,
03:55
but by mapping them on patterns in higher-dimensional space.
77
235583
4435
而是将它们映射到高维空间的规律中。
04:00
But for machine learning to work effectively,
78
240042
2267
但是为了让机器学习有效,
04:02
it needs data --
79
242333
1310
它需要数据——
04:03
it needs lots and lots of data.
80
243667
2767
它需要非常、非常多的数据。
04:06
In the past half-century,
81
246458
1726
在过去的半个世纪里,
04:08
marine researchers have painstakingly collected
82
248208
3476
海洋研究人员煞费苦心地收集
04:11
and hand annotated just a few thousand sperm whale vocalizations,
83
251708
4601
并手工注释了几千种抹香鲸的叫声,
04:16
but in order to learn sperm whale communication,
84
256333
2310
但是为了学习抹香鲸的交流,
04:18
we'll need to collect millions,
85
258667
2392
我们需要收集数以百万计、
04:21
if not tens of millions
86
261083
1976
甚至数以千万计
04:23
of carefully annotated sperm whale vocalizations
87
263083
2893
仔细注释的抹香鲸的声音
并和其行为相对应。
04:26
correlated with behaviors.
88
266000
1893
04:27
We'll do it with noninvasive, autonomous, free-swimming robots,
89
267917
3142
我们将使用非侵入式、 自主、自由游泳的机器人、
04:31
aerial-aquatic drones,
90
271083
1685
航空-潜水式无人机、
04:32
bottom-mounted hydrophone arrays
91
272792
1601
海底水听器阵列
04:34
and more.
92
274417
1267
等等。
04:35
We'll work with our close partners at the Dominica Sperm Whale Project
93
275708
3351
我们将与多米尼加 抹香鲸项目的亲密伙伴合作
04:39
to cover a 20-square-kilometer area
94
279083
2101
来覆盖20平方公里,
04:41
that is frequented by over 25 well-known families of sperm whales.
95
281208
4560
那里有超过25个著名的 抹香鲸家族经常出没。
04:45
We're going to put specific focus on the interactions of mothers and calfs,
96
285792
4642
我们将把重点放在 母亲和幼崽之间的互动上,
04:50
training our machine learning algorithms
97
290458
2185
来训练我们机器学习算法
04:52
to learn whale language from the bottom up.
98
292667
2601
从下往上学习鲸鱼的语言。
04:55
All this data we'll have sent through a pipeline
99
295292
2309
所有这些数据都是通过管道传输的,
04:57
and analyzed by the Project CETI translation team.
100
297625
2434
并由CETI项目翻译团队进行分析。
05:00
The raw audio and context data will go through our machine learning engine
101
300083
3976
原始音频和环境数据 将通过我们的机器学习引擎,
05:04
where it's going to be first sorted by structure.
102
304083
2351
在那里它首先要被按结构排序。
05:06
The linguistics team will then search for things like syntax
103
306458
2893
然后语言学团队 将搜索诸如语法之类的东西
05:09
and time displacement.
104
309375
1268
和时间位移。
05:10
For example,
105
310667
1267
例如,
05:11
if we find an event where a whale was talking about something yesterday,
106
311958
3810
如果我们发现一条鲸鱼 在讨论着昨天发生的事情,
05:15
that alone would be a major finding,
107
315792
2142
光是这个就能成为一个重大发现了,
05:17
something that has thus far only been shown in humans.
108
317958
3268
那是目前为止只在人类中发生的事情。
05:21
And once we've really mastered listening,
109
321250
2309
一旦我们真正掌握了聆听,
05:23
we're going to try really carefully to talk back
110
323583
3226
我们会尝试很小心地回应,
05:26
even on the most simplistic level.
111
326833
2185
即使在最简单的层面上。
05:29
Finally, Project CETI will build an open-source platform
112
329042
2726
最后,CETI项目 将构建一个开源平台,
05:31
where we will make our data sets available to the public,
113
331792
2809
我们将在那向公众提供我们的数据集,
05:34
encouraging the global community
114
334625
2059
鼓励全球社会
05:36
to come along on this journey for understanding.
115
336708
2393
都参与进这段理解的旅程中。
05:39
These animals could be the most intelligent beings on this planet.
116
339125
4393
这些动物可能是 这个星球上最聪明的生物。
05:43
They have a neocortex and spindle cells --
117
343542
2851
它们有一个新皮层和梭形细胞——
05:46
structure that in humans control our higher thoughts,
118
346417
3101
那是在人类身上控制高级思想的结构,
05:49
emotions, memory, language and love.
119
349542
3226
包括情感、记忆、语言和爱。
05:52
And all the platforms that we develop can be cross-applied to other animals:
120
352792
3684
我们开发的所有平台 都可以交叉应用于其他动物:
05:56
to elephants, birds,
121
356500
1601
在大象、鸟类、
05:58
primates, dolphins --
122
358125
1643
灵长类、海豚——
05:59
essentially any animal.
123
359792
1517
几乎所有动物身上。
06:01
In the late 1960s,
124
361333
1643
在20世纪60年代末,
我们的团队成员 罗杰·佩恩发现鲸鱼会唱歌。
06:03
our team member, Roger Payne, discovered that whales sing.
125
363000
3976
(录音:鲸鱼歌唱)
06:07
(Recording: whale singing)
126
367000
1268
06:08
A finding that sparked the Save the Whales movement
127
368292
2642
这一发现引发了拯救鲸鱼运动,
06:10
led to the end of large-scale whaling
128
370958
2643
终结了大规模的捕鲸,
06:13
and prevented several whale species from extinction
129
373625
3476
并阻止了一些鲸鱼物种的灭绝。
06:17
just by showing that whales sing.
130
377125
3143
这光是展示鲸鱼能唱歌就能做到的。
06:20
Imagine if we could understand what they're saying.
131
380292
2476
想象一下如果我们能理解 它们在说什么。
06:22
Now is the time to open this larger dialogue.
132
382792
3642
现在是开启这一更大对话的时候了。
06:26
Now is the time to listen deeply
133
386458
2643
现在是时候用心聆听
06:29
and show these magical animals,
134
389125
1893
来向这些神奇的动物,
06:31
and each other,
135
391042
1851
和彼此,
06:32
newfound respect.
136
392917
1375
展示新建立的尊重。
06:35
Thank you.
137
395417
1250
谢谢。
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