ABOUT THE SPEAKER
Tom Griffiths - Psychologist, cognitive scientist
Tom Griffiths uses ideas from computer science to understand how human minds work.

Why you should listen

Tom Griffiths's research explores connections between natural and artificial intelligence to discover how people solve the challenging computational problems they encounter in everyday life. Currently the Henry R. Luce Professor of Information Technology, Consciousness, and Culture at Princeton University, his work has received awards from organizations ranging from the American Psychological Association to the Sloan Foundation.

In 2016, Griffiths and his friend and collaborator Brian Christian published Algorithms to Live By, a book that illustrates how understanding the algorithms used by computers can inform human decision-making (and vice versa). The book was named one of the Amazon.com "Best Science Books of 2016" and appeared on Forbes's "Must-read brain books of 2016" list as well as the MIT Technology Review's "Best books of 2016" list.

More profile about the speaker
Tom Griffiths | Speaker | TED.com
TEDxSydney

Tom Griffiths: 3 ways to make better decisions -- by thinking like a computer

湯姆葛里菲斯: 做出更好決策的三種方式——採用跟電腦一樣的思考方式

Filmed:
3,652,976 views

如果你曾經為了做決策而掙扎,不妨看看這場演說。認知科學家湯姆葛里菲斯說明我們要如何應用電腦的邏輯來處理難搞的人類問題。他分享三種實際的策略,協助做出更好的決策,適用於各種問題,從找房子到選擇晚餐要去哪一家餐廳都可以用得上。
- Psychologist, cognitive scientist
Tom Griffiths uses ideas from computer science to understand how human minds work. Full bio

Double-click the English transcript below to play the video.

00:13
If there's one city in the world世界
0
1407
1581
如果世界上有一個城市
00:15
where it's hard to find
a place地點 to buy購買 or rent出租,
1
3012
2333
很難找到出售或是出租的地方,
00:17
it's Sydney悉尼.
2
5369
1150
那就是雪梨。
00:19
And if you've tried試著
to find a home here recently最近,
3
7043
2367
如果你最近試著在這裡找個家,
00:21
you're familiar with the problem問題.
4
9434
1840
你對這個問題就會很熟悉。
00:23
Every一切 time you walk步行 into an open打開 house,
5
11298
2014
每當你走進開放看屋的地點,
00:25
you get some information信息
about what's out there
6
13336
2191
你就可以得到些資訊,
知道那裡有什麼,
00:27
and what's on the market市場,
7
15551
1393
以及市場上有什麼;
00:28
but every一切 time you walk步行 out,
8
16968
1430
但每當你走出來時,
00:30
you're running賽跑 the risk風險
of the very best最好 place地點 passing通過 you by.
9
18422
3214
你就冒著錯過最佳選擇的風險。
00:33
So how do you know when
to switch開關 from looking
10
21660
2820
所以,你怎麼知道
何時要從「看看」切換成
00:36
to being存在 ready準備 to make an offer提供?
11
24504
1562
準備好提出交易條件?
00:39
This is such這樣 a cruel殘忍 and familiar problem問題
12
27663
2401
這是個殘酷又熟悉的問題,
00:42
that it might威力 come as a surprise
that it has a simple簡單 solution.
13
30088
3074
讓人意外的是,
它的解決方案很簡單。
00:45
37 percent百分.
14
33588
1203
37%。
00:46
(Laughter笑聲)
15
34815
1828
(笑聲)
00:48
If you want to maximize最大化 the probability可能性
that you find the very best最好 place地點,
16
36667
3936
如果你想要把找到
最佳選擇的機率提升到最高,
00:52
you should look at 37 percent百分
of what's on the market市場,
17
40627
2825
你得要看過市場上
37% 的所有選擇的,
00:55
and then make an offer提供
on the next下一個 place地點 you see,
18
43476
2310
接著到下一個地方時,
就提出交易條件,
00:57
which哪一個 is better than anything
that you've seen看到 so far.
19
45810
2524
它會比你目前看過的
所有選擇都更好。
01:00
Or if you're looking for a month,
take 37 percent百分 of that time --
20
48358
3805
或者,如果你要花一個月來尋找,
就取那段時間的 37% ——
01:04
11 days, to set a standard標準 --
21
52187
2915
即 11 天,來設定標準——
接著你就可以準備行動了。
01:07
and then you're ready準備 to act法案.
22
55126
1575
01:09
We know this because
trying to find a place地點 to live生活
23
57829
2709
我們知道要這麼做,
是因為試圖找住房
01:12
is an example of an optimal最佳
stopping停止 problem問題.
24
60562
2325
就是「最佳停止問題」的例子。
01:14
A class of problems問題 that has been
studied研究 extensively廣泛
25
62911
2483
這類問題一直被數學家
01:17
by mathematicians數學家 and computer電腦 scientists科學家們.
26
65418
2105
和電腦科學家廣為研究。
01:21
I'm a computational計算 cognitive認知 scientist科學家.
27
69502
2519
我是一位計算認知科學家。
01:24
I spend my time trying to understand理解
28
72045
1960
我把時間花在了解
01:26
how it is that human人的 minds頭腦 work,
29
74029
1798
人類大腦如何運作,
01:27
from our amazing驚人 successes成功
to our dismal慘淡 failures故障.
30
75851
3671
從達成了不起的成功
到遭遇令人沮喪的失敗。
01:32
To do that, I think about
the computational計算 structure結構體
31
80552
2493
要做到這一點,我得要思考
日常問題的計算結構,
01:35
of the problems問題
that arise出現 in everyday每天 life,
32
83069
2659
01:37
and compare比較 the ideal理想
solutions解決方案 to those problems問題
33
85752
2375
並將那些問題的理想解決方案
與我們的真實行為做比較。
01:40
to the way that we actually其實 behave表現.
34
88151
1747
01:42
As a side effect影響,
35
90725
1185
它有一個副作用,
01:43
I get to see how applying應用
a little bit of computer電腦 science科學
36
91934
2715
我可以看到應用一點點電腦科學
01:46
can make human人的 decision-making做決定 easier更輕鬆.
37
94673
1860
如何能讓人類決策變得更容易。
01:49
I have a personal個人 motivation動機 for this.
38
97828
1847
我這麼做,背後有個私人的動機。
01:52
Growing生長 up in Perth珀斯
as an overly過於 cerebral顱內 kid孩子 ...
39
100153
3166
我在伯斯長大,以前
是個過度理智的小孩……
01:55
(Laughter笑聲)
40
103343
4731
(笑聲)
02:00
I would always try and act法案 in the way
that I thought was rational合理的,
41
108098
3147
我總是試著用我認為
合理的方式來做事,
02:03
reasoning推理 through通過 every一切 decision決定,
42
111269
1597
做每個決策都要依理推論,
02:04
trying to figure數字 out
the very best最好 action行動 to take.
43
112890
2403
試圖找出採取哪種做法最理想。
02:07
But this is an approach途徑
that doesn't scale規模 up
44
115972
2190
但這種方法無法做更廣的應用,
02:10
when you start開始 to run into
the sorts排序 of problems問題
45
118186
2242
當你開始遇到成人
生活中的那些問題時,
02:12
that arise出現 in adult成人 life.
46
120452
1500
就派不上用場了。
02:13
At one point, I even tried試著
to break打破 up with my girlfriend女朋友
47
121976
2728
我有一度甚至打算要和女友分手,
02:16
because trying to take into account帳戶
her preferences優先 as well as my own擁有
48
124728
3350
原因是我試著考量
她的偏好和我的偏好,
02:20
and then find perfect完善 solutions解決方案 --
49
128102
1683
以找出最完美的解決方案——
02:21
(Laughter笑聲)
50
129809
2259
(笑聲)
02:24
was just leaving離開 me exhausted.
51
132092
1752
我真的被搞得疲憊不堪。
02:25
(Laughter笑聲)
52
133868
2533
(笑聲)
02:28
She pointed out that I was taking服用
the wrong錯誤 approach途徑
53
136425
2429
她指出我在解決這個問題時
02:30
to solving this problem問題 --
54
138878
1609
用錯了方法——
02:32
and she later後來 became成為 my wife妻子.
55
140511
1382
後來她成了我的太太。
02:33
(Laughter笑聲)
56
141917
2062
(笑聲)
02:36
(Applause掌聲)
57
144003
4971
(掌聲)
02:40
Whether是否 it's as basic基本 as trying to decide決定
what restaurant餐廳 to go to
58
148998
3461
不論是很基本的問題,
比如決定要去哪家餐廳吃飯,
02:44
or as important重要 as trying to decide決定
who to spend the rest休息 of your life with,
59
152483
4055
或是很重要的問題,
比如決定要和誰共渡餘生,
02:48
human人的 lives生活 are filled填充
with computational計算 problems問題
60
156562
2375
人生其實都充滿了計算問題,
02:50
that are just too hard to solve解決
by applying應用 sheer絕對 effort功夫.
61
158961
4196
光靠努力是很難解決的。
02:55
For those problems問題,
62
163650
1166
那些問題
02:56
it's worth價值 consulting諮詢 the experts專家:
63
164840
1961
值得去諮詢專家:
02:58
computer電腦 scientists科學家們.
64
166825
1151
電腦科學家。
03:00
(Laughter笑聲)
65
168000
1784
(笑聲)
03:01
When you're looking for life advice忠告,
66
169808
1819
當你要尋求人生忠告時,
03:03
computer電腦 scientists科學家們 probably大概 aren't
the first people you think to talk to.
67
171651
3640
你最先想要問的人大概
不會是電腦科學家。
03:07
Living活的 life like a computer電腦 --
68
175315
1875
把人生過得像電腦一樣——
03:09
stereotypically刻板印象 deterministic確定性,
exhaustive詳細 and exact精確 --
69
177214
2578
刻板的決定論、
詳盡無遺,且精確——
03:11
doesn't sound聲音 like a lot of fun開玩笑.
70
179816
1552
聽起來實在不好玩。
03:14
But thinking思維 about the computer電腦 science科學
of human人的 decisions決定
71
182153
2927
但思考一下人類決策的電腦科學,
03:17
reveals揭示 that in fact事實,
we've我們已經 got this backwards向後.
72
185104
2408
會發現,事實上,
我們把方向弄反了。
03:19
When applied應用的 to the sorts排序
of difficult problems問題
73
187536
2198
當應用在人生中的
那些困難問題上時,
03:21
that arise出現 in human人的 lives生活,
74
189758
1239
03:23
the way that computers電腦
actually其實 solve解決 those problems問題
75
191021
2727
電腦實際上用來解決
那些問題的方式
03:25
looks容貌 a lot more like the way
that people really act法案.
76
193772
2896
看起來很像是人們真正使用的方式。
03:29
Take the example of trying to decide決定
what restaurant餐廳 to go to.
77
197267
2922
就用決定要去哪間餐廳
吃飯當作例子吧。
03:33
This is a problem問題 that has
a particular特定 computational計算 structure結構體.
78
201090
3088
這個問題有特定的計算結構。
03:36
You've got a set of options選項,
79
204202
1524
你有一組選項,
03:37
you're going to choose選擇
one of those options選項,
80
205750
2049
你得要從那些選項中擇一,
03:39
and you're going to face面對
exactly究竟 the same相同 decision決定 tomorrow明天.
81
207823
2811
且你明天還會面對
完全一樣的決策。
03:42
In that situation情況,
82
210658
1151
在那樣的情況下,
03:43
you run up against反對
what computer電腦 scientists科學家們 call
83
211833
2295
你碰到的就是電腦科學家所謂的
03:46
the "explore-exploit探索-開發 trade-off交易."
84
214152
2372
「探索/利用的權衡」。
03:49
You have to make a decision決定
85
217115
1334
你得要做一個決策,
03:50
about whether是否 you're going
to try something new --
86
218473
2334
決定你是否要嘗試新選項——
03:52
exploring探索, gathering蒐集 some information信息
87
220831
2183
去「探索」,收集一些未來
03:55
that you might威力 be able能夠
to use in the future未來 --
88
223038
2285
可能會用到的資訊——
03:57
or whether是否 you're going to go to a place地點
that you already已經 know is pretty漂亮 good --
89
225347
3793
或者你是否要選擇去
你已經知道不錯的地方——
04:01
exploiting利用 the information信息
that you've already已經 gathered雲集 so far.
90
229164
3003
「利用」你目前已經
收集到的資訊。
04:05
The explore探索/exploit利用 trade-off交易
shows節目 up any time you have to choose選擇
91
233488
3096
探索/利用的權衡會出現在每次
04:08
between之間 trying something new
92
236608
1350
你必須要從新選項和已經知道
不錯的選項中擇一的情況下,
04:09
and going with something
that you already已經 know is pretty漂亮 good,
93
237982
2980
也許是聽音樂,
04:12
whether是否 it's listening to music音樂
94
240986
1550
04:14
or trying to decide決定
who you're going to spend time with.
95
242560
2634
或者是試著決定
你要跟誰一起殺時間。
04:17
It's also the problem問題
that technology技術 companies公司 face面對
96
245218
2494
這也是科技公司會面臨的問題,
比如決定要在網頁上放什麼
廣告時,遇到的就是這種問題。
04:19
when they're trying to do something
like decide決定 what ad廣告 to show顯示 on a web捲筒紙 page.
97
247736
3667
它們應該要刊登新廣告,
從中得到一些資訊嗎?
04:23
Should they show顯示 a new ad廣告
and learn學習 something about it,
98
251417
2572
或是它們應該要給你看
04:26
or should they show顯示 you an ad廣告
99
254003
1383
一則它們已經知道你很有可能
會點選的廣告?
04:27
that they already已經 know there's a good
chance機會 you're going to click點擊 on?
100
255420
3306
在過去六十年,
04:30
Over the last 60 years年份,
101
258750
1151
電腦科學家在了解
探索/利用的權衡上,
04:31
computer電腦 scientists科學家們 have made製作
a lot of progress進展 understanding理解
102
259925
2892
04:34
the explore探索/exploit利用 trade-off交易,
103
262841
1452
有相當多進展,
他們的結果帶來了
一些讓人吃驚的洞見。
04:36
and their results結果 offer提供
some surprising奇怪 insights見解.
104
264317
2398
04:39
When you're trying to decide決定
what restaurant餐廳 to go to,
105
267291
2598
當你要試著決定該去哪一間餐廳時,
04:41
the first question you should ask yourself你自己
106
269913
2000
你應該先問你自己一個問題:
04:43
is how much longer
you're going to be in town.
107
271937
2179
你還會待在鎮上多久?
04:46
If you're just going to be there
for a short time,
108
274505
2342
如果你只是短暫停留,
04:48
then you should exploit利用.
109
276871
1515
那麼你應該要「利用」。
04:50
There's no point gathering蒐集 information信息.
110
278410
1905
收集資訊是沒有意義的。
04:52
Just go to a place地點
you already已經 know is good.
111
280339
2048
直接去一個你已經
知道不錯的地方吧。
04:54
But if you're going to be there
for a longer time, explore探索.
112
282411
2929
但如果你會待久一點,
就「探索」吧。
04:57
Try something new,
because the information信息 you get
113
285364
2387
試試新選項,因為
你從中得到的資訊
04:59
is something that can improve提高
your choices選擇 in the future未來.
114
287775
2974
可能協助你在未來做更好的選擇。
05:02
The value of information信息 increases增加
115
290773
1979
你越有可能用到一項資訊,
05:04
the more opportunities機會
you're going to have to use it.
116
292776
2600
該資訊的價值就會增加。
05:08
This principle原理 can give us insight眼光
117
296193
1754
這條原則也能協助我們
05:09
into the structure結構體
of a human人的 life as well.
118
297971
2070
洞察人類的人生。
05:13
Babies嬰兒 don't have a reputation聲譽
for being存在 particularly尤其 rational合理的.
119
301104
3053
寶寶通常不會特別理性。
05:17
They're always trying new things,
120
305098
1825
他們總是在嘗試新東西,
05:18
and you know, trying to stick them
in their mouths嘴巴.
121
306947
2836
你們知道的,總把
新東西放到嘴巴裡。
05:22
But in fact事實, this is exactly究竟
what they should be doing.
122
310621
2952
但,事實上,他們
的確應該要這麼做。
05:25
They're in the explore探索
phase of their lives生活,
123
313597
2430
他們正處在人生的探索階段,
05:28
and some of those things
could turn out to be delicious美味的.
124
316051
2621
他們嘗試的東西當中,
有些可能真的會很美味。
05:32
At the other end結束 of the spectrum光譜,
125
320040
1572
在光譜的另一端,
05:33
the old guy who always goes
to the same相同 restaurant餐廳
126
321636
2506
是老人,他們總是去同樣的餐廳,
05:36
and always eats the same相同 thing
127
324166
1635
總是點同樣的食物,
05:37
isn't boring無聊 --
128
325825
1328
並不是無趣,
05:39
he's optimal最佳.
129
327177
1509
而是最佳化的選擇。
05:40
(Laughter笑聲)
130
328710
3830
(笑聲)
05:44
He's exploiting利用 the knowledge知識
that he's earned
131
332564
2228
他在利用他從一生的經驗中
05:46
through通過 a lifetime's一輩子的 experience經驗.
132
334816
1767
已經得到的知識。
05:50
More generally通常,
133
338406
1151
更普遍來說,知道有
「探索/利用的權衡」,
05:51
knowing會心 about
the explore探索/exploit利用 trade-off交易
134
339581
2031
就能讓你在做決策時能更輕鬆些,
05:53
can make it a little easier更輕鬆 for you
to sort分類 of relax放鬆 and go easier更輕鬆 on yourself你自己
135
341636
3667
不要對自己太嚴厲。
05:57
when you're trying to make a decision決定.
136
345327
1810
你不需要每晚都去最好的餐廳。
05:59
You don't have to go
to the best最好 restaurant餐廳 every一切 night.
137
347161
2728
06:01
Take a chance機會, try something new, explore探索.
138
349913
2979
冒個險,嘗試新餐廳,去探索。
06:04
You might威力 learn學習 something.
139
352916
1627
你可能會學到些什麼。
06:06
And the information信息 that you gain獲得
140
354567
1589
而你所得到的資訊
06:08
is going to be worth價值 more
than one pretty漂亮 good dinner晚餐.
141
356180
2543
價值絕對勝過一頓好吃的晚餐。
06:12
Computer電腦 science科學 can also help
to make it easier更輕鬆 on us
142
360178
2699
在家中或在辦公室裡的其他地方,
電腦科學也能夠讓我們更輕鬆些。
06:14
in other places地方 at home and in the office辦公室.
143
362901
2152
06:17
If you've ever had
to tidy整潔 up your wardrobe衣櫃,
144
365860
2453
如果你得要整理你的衣櫥,
06:20
you've run into a particularly尤其
agonizing折騰 decision決定:
145
368337
2671
你會碰到一個特別煩惱的決定:
06:23
you have to decide決定 what things
you're going to keep
146
371032
2382
你得要決定哪些東西該留下,
06:25
and what things you're going to give away.
147
373438
2009
哪些東西該送人。
06:27
Martha瑪莎 Stewart斯圖爾特 turns out
to have thought very hard about this --
148
375974
3000
結果發現瑪莎史都華花了
很多功夫在想這件事——
06:30
(Laughter笑聲)
149
378998
1205
(笑聲)
06:32
and she has some good advice忠告.
150
380227
1678
她有些不錯的忠告。
06:33
She says, "Ask yourself你自己 four questions問題:
151
381929
2295
她說:「問你自己四個問題:
06:36
How long have I had it?
152
384248
1526
我已經持有它多久了?
06:37
Does it still function功能?
153
385798
1450
它還有功能嗎?
06:39
Is it a duplicate重複
of something that I already已經 own擁有?
154
387272
3199
它是不是跟某樣
我已經擁有的東西一樣?
06:42
And when was the last time
I wore穿著 it or used it?"
155
390495
2411
我上次穿它或用它是什麼時候?」
06:46
But there's another另一個 group of experts專家
156
394648
1715
但還有另一群專家
06:48
who perhaps也許 thought
even harder更難 about this problem問題,
157
396387
3130
花了更多功夫在想這個問題,
06:51
and they would say one of these questions問題
is more important重要 than the others其他.
158
399541
3660
他們會說,這些問題當中
有一個比其他的都還重要。
06:55
Those experts專家?
159
403819
1150
那些專家是誰?
06:57
The people who design設計
the memory記憶 systems系統 of computers電腦.
160
405281
2848
設計出電腦記憶體系統的人。
07:00
Most computers電腦 have
two kinds of memory記憶 systems系統:
161
408153
2254
大部分的電腦有兩種記憶體系統:
07:02
a fast快速 memory記憶 system系統,
162
410431
1387
快速記憶體系統,
07:03
like a set of memory記憶 chips芯片
that has limited有限 capacity容量,
163
411842
3169
就像是一組記憶體晶片,容量有限,
07:07
because those chips芯片 are expensive昂貴,
164
415035
2052
因為那些晶片很貴,
07:09
and a slow memory記憶 system系統,
which哪一個 is much larger.
165
417111
3260
還有慢速記憶體系統,
它的容量大很多。
07:13
In order訂購 for the computer電腦 to operate操作
as efficiently有效率的 as possible可能,
166
421239
3000
為了要讓電腦的
運作效能盡可能提高,
07:16
you want to make sure
167
424263
1151
你會希望能確保你要存取的資訊
07:17
that the pieces of information信息
you want to access訪問
168
425438
2352
位在快速記憶體系統中,
這樣你就能快速取得它。
07:19
are in the fast快速 memory記憶 system系統,
169
427814
1477
07:21
so that you can get to them quickly很快.
170
429315
1754
每當你存取一項資訊時,
07:23
Each time you access訪問
a piece of information信息,
171
431093
2061
它就會被載入快速記憶體中,
07:25
it's loaded into the fast快速 memory記憶
172
433178
1525
電腦得要決定要從
快速記憶體中移除哪個項目,
07:26
and the computer電腦 has to decide決定 which哪一個 item項目
it has to remove去掉 from that memory記憶,
173
434727
3877
07:30
because it has limited有限 capacity容量.
174
438628
1741
因為它的容量有限。
07:33
Over the years年份,
175
441794
1151
數年來,電腦科學家
試過幾種不同的策略
07:34
computer電腦 scientists科學家們 have tried試著
a few少數 different不同 strategies策略
176
442969
2739
來判定該從快速記憶體中移除什麼。
07:37
for deciding決定 what to remove去掉
from the fast快速 memory記憶.
177
445732
2292
他們有試過隨機選擇的方法,
07:40
They've他們已經 tried試著 things like choosing選擇
something at random隨機
178
448048
3176
07:43
or applying應用 what's called
the "first-in先入, first-out先出 principle原理,"
179
451248
2906
也試過採用「先進先出」的原則,
也就是說把在記憶體當中
最久的項目給移除。
07:46
which哪一個 means手段 removing去除 the item項目
180
454178
1399
07:47
which哪一個 has been in the memory記憶
for the longest最長.
181
455601
2176
07:50
But the strategy戰略 that's most effective有效
182
458091
2713
不過,最有效的策略,
07:52
focuses重點 on the items項目
which哪一個 have been least最小 recently最近 used.
183
460828
3229
是把目標放在近期最少使用的項目。
07:56
This says if you're going to decide決定
to remove去掉 something from memory記憶,
184
464972
3191
這種策略就是,如果你得
從記憶體中移除某樣東西,
08:00
you should take out the thing which哪一個 was
last accessed訪問 the furthest最遠 in the past過去.
185
468187
4968
你應該選擇最後一次使用時間
是最久遠的那樣東西。
08:05
And there's a certain某些
kind of logic邏輯 to this.
186
473179
2159
這背後是有某種邏輯的。
08:07
If it's been a long time since以來 you last
accessed訪問 that piece of information信息,
187
475362
3598
如果你上次存取那項資訊
已經是很久以前的事了,
08:10
it's probably大概 going to be a long time
188
478984
1762
你下次需要存取它的時間
08:12
before you're going to need
to access訪問 it again.
189
480770
2192
應該也會是很久以後。
08:15
Your wardrobe衣櫃 is just like
the computer's電腦 memory記憶.
190
483787
2774
你的衣櫥就像是電腦的記憶體。
08:18
You have limited有限 capacity容量,
191
486585
2030
你的容量有限,
08:20
and you need to try and get in there
the things that you're most likely容易 to need
192
488639
5327
你得要把你最有可能
用到的東西放進去,
08:25
so that you can get to them
as quickly很快 as possible可能.
193
493990
2419
這樣你才能夠盡快取得它們。
認知到這一點後,
08:29
Recognizing認識 that,
194
497016
1151
08:30
maybe it's worth價值 applying應用
the least最小 recently最近 used principle原理
195
498191
2813
也許也值得嘗試應用
「近期最少使用」原則
08:33
to organizing組織 your wardrobe衣櫃 as well.
196
501028
1949
來整理你的衣櫥。
08:35
So if we go back
to Martha's瑪莎 four questions問題,
197
503001
2082
如果我們回到瑪莎的四個問題,
08:37
the computer電腦 scientists科學家們
would say that of these,
198
505107
2861
電腦科學家會說,在這些問題中,
08:39
the last one is the most important重要.
199
507992
1941
最後一個問題是最重要。
08:43
This idea理念 of organizing組織 things
200
511473
2055
在整理東西時,要讓你最可能
08:45
so that the things you are most
likely容易 to need are most accessible無障礙
201
513552
3129
需要的東西最容易存取的這個想法,
08:48
can also be applied應用的 in your office辦公室.
202
516705
1676
也可以應用到你的辦公室中。
08:51
The Japanese日本 economist經濟學家 Yukio鳩山 Noguchi野口
203
519091
1930
日本經濟學家野口悠紀雄
08:53
actually其實 invented發明 a filing備案 system系統
that has exactly究竟 this property屬性.
204
521045
3055
真的發明了一個具有
這種特性的建檔系統。
08:57
He started開始 with a cardboard紙板 box,
205
525161
1609
他從一個紙箱子開始,
08:58
and he put his documents文件 into the box
from the left-hand左手 side.
206
526794
2940
他把他的文件
從左到右放進箱子中。
09:02
Each time he'd他會 add a document文件,
207
530067
1434
每當他放入一份文件時,
他就得要移動箱中的文件,
09:03
he'd他會 move移動 what was in there along沿
208
531525
1621
才能把新放入的文件
放入箱子的左邊。
09:05
and he'd他會 add that document文件
to the left-hand左手 side of the box.
209
533170
2810
每當他需要使用一份文件時,
他會把該文件取出,
09:08
And each time he accessed訪問
a document文件, he'd他會 take it out,
210
536004
2667
使用完之後放回到最左邊。
09:10
consult請教 it and put it back in
on the left-hand左手 side.
211
538695
2608
09:13
As a result結果, the documents文件 would be
ordered有序 from left to right
212
541327
3281
這樣的結果是,
文件會從左到右排好,
09:16
by how recently最近 they had been used.
213
544632
1761
最左邊的是最近期使用過的。
09:18
And he found發現 he could quickly很快 find
what he was looking for
214
546417
2774
他發現這樣排之後,
他只要從箱子的左邊開始
09:21
by starting開始 at the left-hand左手
side of the box
215
549215
2054
一直向右找,就能快速
找到他想找的文件。
09:23
and working加工 his way to the right.
216
551293
1572
09:25
Before you dash短跑 home
and implement實行 this filing備案 system系統 --
217
553215
2621
在你們衝回家導入
這個建檔系統之前——
09:27
(Laughter笑聲)
218
555860
1731
(笑聲)
09:29
it's worth價值 recognizing認識
that you probably大概 already已經 have.
219
557615
2638
值得先想想,你可能
已經有這個系統了。
09:32
(Laughter笑聲)
220
560891
3310
(笑聲)
09:36
That pile of papers文件 on your desk ...
221
564225
2866
你書桌上的那疊紙……
09:39
typically一般 maligned非議
as messy and disorganized雜亂無章,
222
567115
2474
通常都被別人誹謗說是亂七八糟,
09:41
a pile of papers文件 is, in fact事實,
perfectly完美 organized有組織的 --
223
569613
2565
其實是有著完美
組織系統的一疊紙——
09:44
(Laughter笑聲)
224
572202
1000
(笑聲)
09:45
as long as you, when you take a paper out,
225
573226
2014
只要你每次把一張紙拿出來,
09:47
put it back on the top最佳 of the pile,
226
575264
2432
用完之後會放回那疊紙的最上方,
09:49
then those papers文件 are going
to be ordered有序 from top最佳 to bottom底部
227
577720
2826
那麼那疊紙從上到下
就排好了順序,
09:52
by how recently最近 they were used,
228
580570
1552
最上面的是最近期使用的,
09:54
and you can probably大概 quickly很快 find
what you're looking for
229
582146
2701
你從那疊紙的最上面開始找,
可能就能快速找到你要的。
09:56
by starting開始 at the top最佳 of the pile.
230
584871
1700
09:59
Organizing組織 your wardrobe衣櫃 or your desk
231
587988
1848
整理你的衣櫥或你的書桌
10:01
are probably大概 not the most pressing緊迫
problems問題 in your life.
232
589860
2681
可能不是你人生中最緊迫的問題。
10:05
Sometimes有時 the problems問題 we have to solve解決
are simply只是 very, very hard.
233
593588
3602
有時,我們需要解決的問題
就是非常非常難搞。
但即使在那些情況下,
10:09
But even in those cases,
234
597716
1150
10:10
computer電腦 science科學 can offer提供 some strategies策略
235
598890
2021
電腦科學也能夠提供一些策略,
10:12
and perhaps也許 some solace慰藉.
236
600935
1669
也許還能提供一些安慰。
10:16
The best最好 algorithms算法 are about doing
what makes品牌 the most sense
237
604048
3075
最好的演算法,
就是要在最短的時間內
做出最合理的舉動。
10:19
in the least最小 amount of time.
238
607147
1502
10:22
When computers電腦 face面對 hard problems問題,
239
610627
1958
當電腦面臨困難的問題時,
10:24
they deal合同 with them by making製造 them
into simpler簡單 problems問題 --
240
612609
2762
它們的處理方式是把那些問題
變成更簡單的問題——
做法包括使用隨機性、
10:27
by making製造 use of randomness隨機性,
241
615395
1346
10:28
by removing去除 constraints限制
or by allowing允許 approximations近似值.
242
616765
3817
移除限制式,或是允許近似值。
10:32
Solving解決 those simpler簡單 problems問題
243
620606
1494
解決那些較簡單的問題,
10:34
can give you insight眼光
into the harder更難 problems問題,
244
622124
2920
就能提供你關於
原本困難問題的洞見,
10:37
and sometimes有時 produces產生
pretty漂亮 good solutions解決方案 in their own擁有 right.
245
625068
3031
有時,還能自己產生出
很好的解決方案。
10:41
Knowing會心 all of this has helped幫助 me
to relax放鬆 when I have to make decisions決定.
246
629658
3569
知道這一切,讓我在
必須要做決策時能夠放輕鬆。
10:45
You could take the 37 percent百分 rule規則
for finding發現 a home as an example.
247
633251
3255
可以用找房子時的
37% 規則來當例子。
10:49
There's no way that you can
consider考慮 all of the options選項,
248
637015
2702
你不可能把所有的
選項都納入考量,
10:51
so you have to take a chance機會.
249
639741
1580
所以你得要冒險。
10:53
And even if you follow跟隨
the optimal最佳 strategy戰略,
250
641888
2860
即使你遵循最佳化策略,
10:56
you're not guaranteed保證 a perfect完善 outcome結果.
251
644772
1931
也不能保證你會得到最完美的結果。
10:59
If you follow跟隨 the 37 percent百分 rule規則,
252
647198
2154
如果你遵循 37% 規則,
11:01
the probability可能性 that you find
the very best最好 place地點 is --
253
649376
3279
你能找到最棒的地方的機率是——
11:04
funnily好笑 enough足夠 ...
254
652679
1508
很有趣……
11:06
(Laughter笑聲)
255
654211
1014
(笑聲)
11:07
37 percent百分.
256
655249
1420
是 37%。
11:09
You fail失敗 most of the time.
257
657966
1689
大部分的時候,你會失敗。
11:12
But that's the best最好 that you can do.
258
660522
1744
但你能做到最好的就是這樣了。
11:14
Ultimately最終,, computer電腦 science科學
can help to make us more forgiving寬容
259
662846
3052
最終,電腦科學會協助讓我們
更能原諒自己的限制。
11:17
of our own擁有 limitations限制.
260
665922
1909
11:20
You can't control控制 outcomes結果,
just processes流程.
261
668333
2539
你不能控制結果,只能控制過程。
11:22
And as long as you've used
the best最好 process處理,
262
670896
2119
只要你已經用了最好的過程,
11:25
you've doneDONE the best最好 that you can.
263
673039
1763
你就已經盡了全力。
11:26
Sometimes有時 those best最好 processes流程
involve涉及 taking服用 a chance機會 --
264
674826
3688
有時,最好的過程會需要冒點險——
11:30
not considering考慮 all of your options選項,
265
678538
2115
比如不去考量所有的選項,
11:32
or being存在 willing願意 to settle解決
for a pretty漂亮 good solution.
266
680677
2729
或是願意妥協,接受
算是不錯的解決方案。
11:35
These aren't the concessions讓步
that we make when we can't be rational合理的 --
267
683430
3350
這些並不是我們在無法
理性時所做的讓步——
11:38
they're what being存在 rational合理的 means手段.
268
686804
1629
它們就是理性的真締。
11:40
Thank you.
269
688833
1151
謝謝大家。
11:42
(Applause掌聲)
270
690008
4547
(掌聲)
Translated by Lilian Chiu
Reviewed by Helen Chang

▲Back to top

ABOUT THE SPEAKER
Tom Griffiths - Psychologist, cognitive scientist
Tom Griffiths uses ideas from computer science to understand how human minds work.

Why you should listen

Tom Griffiths's research explores connections between natural and artificial intelligence to discover how people solve the challenging computational problems they encounter in everyday life. Currently the Henry R. Luce Professor of Information Technology, Consciousness, and Culture at Princeton University, his work has received awards from organizations ranging from the American Psychological Association to the Sloan Foundation.

In 2016, Griffiths and his friend and collaborator Brian Christian published Algorithms to Live By, a book that illustrates how understanding the algorithms used by computers can inform human decision-making (and vice versa). The book was named one of the Amazon.com "Best Science Books of 2016" and appeared on Forbes's "Must-read brain books of 2016" list as well as the MIT Technology Review's "Best books of 2016" list.

More profile about the speaker
Tom Griffiths | Speaker | TED.com