Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:02,240 --> 00:00:04,520
So we've just seen a short history of AI.
2
00:00:04,520 --> 00:00:09,989
and now the question is where are we now.
3
00:00:09,989 --> 00:00:11,459
We'll, let's let you vote.
4
00:00:11,459 --> 00:00:14,599
We'll do a little quiz. Which of the following things
5
00:00:14,599 --> 00:00:18,410
can we do at present.
6
00:00:18,410 --> 00:00:22,990
Can we create a machine which plays a
decent game of table tennis?
7
00:00:22,990 --> 00:00:26,230
Raise your hands if yes.
8
00:00:26,230 --> 00:00:27,710
Yes.
9
00:00:27,710 --> 00:00:33,100
Can we create a game that plays a decent
game of Jeopardy?
10
00:00:33,100 --> 00:00:37,450
Yes; many of you saw this decent game--it did okay, right, computer did okay.
11
00:00:37,450 --> 00:00:40,530
Okay, we'll actually talk more about Watson later.
12
00:00:40,530 --> 00:00:42,530
Can we build an AI,
13
00:00:42,530 --> 00:00:48,110
that will drive safely along a curving
mountain road? Yes.
14
00:00:48,110 --> 00:00:53,700
Can we build an AI that'll drive safely
along Telegraph Avenue?
15
00:00:53,700 --> 00:00:57,010
I'm not sure *I* can drive safely on
Telegraph Avenue. So the answer is
16
00:00:57,010 --> 00:00:59,760
actually tricky. We'll come back to
autonomous driving
17
00:00:59,760 --> 00:01:00,950
a couple times in this course.
18
00:01:00,950 --> 00:01:04,500
I'm going to put a question mark here--you know
Telegraph Avenue is particularly difficult
19
00:01:04,500 --> 00:01:08,980
and we can't just have a a car which
does everything a human can do.
20
00:01:08,980 --> 00:01:11,160
However, there are autonomous cars out there driving.
21
00:01:11,160 --> 00:01:14,750
They are among us and they have a plan.
In fact we'll talk a little bit about for example
22
00:01:14,750 --> 00:01:18,540
the Google autonomous driving cars, which
under the right circumstances and with
23
00:01:18,540 --> 00:01:23,840
tools that a driver does not have, are in
fact able to drive on real roads. Amazing progress.
24
00:01:23,840 --> 00:01:28,460
One thing that's really cool about this list:
I've been using this list for years, and
25
00:01:28,460 --> 00:01:30,499
kind of every year or two
26
00:01:30,499 --> 00:01:33,429
some "X" turns into a question mark or a
question mark turns into a check.
27
00:01:33,429 --> 00:01:35,090
It's really cool to watch.
28
00:01:35,090 --> 00:01:40,840
Can we build an AI that will buy a week's worth
of groceries on the web.
29
00:01:40,840 --> 00:01:44,360
Yeah, it's kind of almost not even AI right...
30
00:01:44,360 --> 00:01:47,100
Can we build an AI which will buy a
week's worth of groceries at the Berkeley Bowl?
31
00:01:47,100 --> 00:01:49,320
32
00:01:49,320 --> 00:01:53,240
Well that's a little harder right--you've gotta like navigate around carts.
33
00:01:53,240 --> 00:01:57,060
This is probably something that we currently cannot do and this is an
illustration of the decision problem--
34
00:01:57,060 --> 00:01:58,270
What do I buy?
35
00:01:58,270 --> 00:02:02,140
--being tangled up in the action in the
real world which involves like,
36
00:02:02,140 --> 00:02:03,730
not bumping into people with your cart.
37
00:02:03,730 --> 00:02:07,690
To operate in the real world and in an
actual mobile platform often mixes all
38
00:02:07,690 --> 00:02:10,910
this stuff together in a very challenging
way where the decisions you make happen
39
00:02:10,910 --> 00:02:13,459
at multiple levels of abstraction.
40
00:02:13,459 --> 00:02:17,489
Can we build an AI that will discover and
prove a new mathematical theorem.
41
00:02:17,489 --> 00:02:19,340
Raise your hand if yes.
42
00:02:19,340 --> 00:02:21,199
Raise your hand if no.
43
00:02:21,199 --> 00:02:25,929
Okay, so I'm gonna put a question mark. We
can in fact build amazing theorem provers
44
00:02:25,929 --> 00:02:27,819
that can prove new theorems.
45
00:02:27,819 --> 00:02:31,949
Discovery, not so much. Why is this hard? So it turns out that if
46
00:02:31,949 --> 00:02:33,369
I state a theorem and it's true,
47
00:02:33,369 --> 00:02:36,709
we can build automated methods of
proving that they are true through computation.
48
00:02:36,709 --> 00:02:38,119
49
00:02:38,119 --> 00:02:41,639
It's a lot harder to just ask the computer:
Tell me something awesome about algebra I don't know.
50
00:02:41,639 --> 00:02:43,039
51
00:02:43,039 --> 00:02:45,429
Cause it could be like, hey, wait, wait, I got it.
52
00:02:45,429 --> 00:02:47,239
10 + 3 = 13.
53
00:02:47,239 --> 00:02:49,950
You know, it's true! But you don't
know what's interesting so it's very
54
00:02:49,950 --> 00:02:52,370
hard to characterize what makes a theorem interesting.
55
00:02:52,370 --> 00:02:55,020
In some sense it's easier to prove
something that's true
56
00:02:55,020 --> 00:02:56,849
than figure out what should be proven.
57
00:02:56,849 --> 00:03:00,029
Can we build an AI that will converse
successfully with another person for an hour?
58
00:03:00,029 --> 00:03:02,709
59
00:03:02,709 --> 00:03:07,819
It depends a lot on the other person.
60
00:03:07,819 --> 00:03:10,510
Some of you may have heard of ELIZA. We'll talk a little bit about ELIZA when we
61
00:03:10,510 --> 00:03:14,510
talk about natural language later in the
course, but, for most people the answer is
62
00:03:14,510 --> 00:03:15,189
no.
63
00:03:15,189 --> 00:03:17,999
For some people the answer is yes--you
can converse with some of the people for
64
00:03:17,999 --> 00:03:21,409
some of the time, I don't know, but the
point is in general we cannot yet do this.
65
00:03:21,409 --> 00:03:24,919
We cannot yet have computers that
have full conversational ability on any topic
66
00:03:24,919 --> 00:03:26,859
just like a human would.
67
00:03:26,859 --> 00:03:30,039
How about we build in an AI that can
perform a surgical operation?
68
00:03:30,039 --> 00:03:31,359
Who says yes.
69
00:03:31,359 --> 00:03:33,539
Okay, raise your hand if you would let
and an AI perform a surgical operaton on you.
70
00:03:33,539 --> 00:03:34,909
71
00:03:34,909 --> 00:03:37,359
Would you let an AI remove your
kidney?
72
00:03:37,359 --> 00:03:38,929
Right, okay,
73
00:03:38,929 --> 00:03:41,429
where did all the hands go?
74
00:03:41,429 --> 00:03:44,939
That illustrates one of the issues--the
answer here's a little complicated.
75
00:03:44,939 --> 00:03:47,290
Yeah, we have computer-assisted surgery.
76
00:03:47,290 --> 00:03:50,800
Computers are starting--including some cool
stuff in Pieter Abbeel's group--starting to
77
00:03:50,800 --> 00:03:54,470
be able to tie knots in automatic way.
You can do things like track the
78
00:03:54,470 --> 00:03:58,079
movement of an eye in eye surgery and things like that.
79
00:03:58,079 --> 00:03:59,619
However,
80
00:03:59,619 --> 00:04:03,389
it is not yet the case, both in
technology and in cultural kind of
81
00:04:03,389 --> 00:04:06,709
willingness to undergo the experiment,
that we're gonna be having computers
82
00:04:06,709 --> 00:04:09,239
swap our kidneys around.
83
00:04:09,239 --> 00:04:13,889
Can we build an AI that can put away the dishes and fold the laundry?
84
00:04:13,889 --> 00:04:17,650
Yeah, we can build it right here at Berkeley.
I can show you a little bit later, again, some of
85
00:04:17,650 --> 00:04:20,159
Pieter's work on folding.
86
00:04:20,159 --> 00:04:21,699
We can do that.
87
00:04:21,699 --> 00:04:26,800
Translate spoken Chinese into spoken English in real time?
88
00:04:26,800 --> 00:04:29,610
Raise your hand if you think we can do it.
89
00:04:29,610 --> 00:04:32,479
Yeah, we'll talk more about translation and real time translation later,
90
00:04:32,479 --> 00:04:33,529
but we can basically do this.
91
00:04:33,529 --> 00:04:37,279
Now you say, wait, is it good English?
92
00:04:37,279 --> 00:04:42,830
Um..., uh..., I didn't say it was good I said it
was spoken right, um...
93
00:04:42,830 --> 00:04:44,710
but we can do real-time
94
00:04:44,710 --> 00:04:47,960
recognition, translation synthesis. That
is something we can do.
95
00:04:47,960 --> 00:04:49,979
Better than you probably think.
96
00:04:49,979 --> 00:04:54,340
Can we build an AI that will write an
intentionally funny story.
97
00:04:54,340 --> 00:04:56,620
Who thinks yes?
98
00:04:56,620 --> 00:04:57,440
Uh...
99
00:04:57,440 --> 00:04:58,909
I'm going to say no.
100
00:04:58,909 --> 00:05:00,870
It's an open area of natural language
processing.
8493
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.
Afrikaans
Akan
Albanian
Amharic
Armenian
Azerbaijani
Basque
Belarusian
Bemba
Bengali
Bihari
Bosnian
Breton
Bulgarian
Cambodian
Catalan
Cebuano
Cherokee
Chichewa
Chinese (Simplified)
Chinese (Traditional)
Corsican
Croatian
Czech
Danish
Dutch
English
Esperanto
Estonian
Faroese
Finnish
Galician
Georgian
German
Greek
Guarani
Haitian Creole
Hausa
Hawaiian
Hebrew
Hmong
Hungarian
Icelandic
Indonesian
Irish
Italian
Japanese
Kazakh
Kinyarwanda
Kirundi
Kongo
Korean
Krio (Sierra Leone)
Kurdish
Kyrgyz
Laothian
Latin
Latvian
Lithuanian
Luganda
Luo
Luxembourgish
Macedonian
Malagasy
Malay
Maltese
Maori
Mauritian Creole
Moldavian
Mongolian
Myanmar (Burmese)
Montenegrin
Nepali
Norwegian
Pashto
Persian
Polish
Portuguese (Brazil)
Portuguese (Portugal)
Quechua
Romanian
Romansh
Russian
Samoan
Scots Gaelic
Serbian
Sesotho
Setswana
Seychellois Creole
Shona
Sindhi
Sinhalese
Slovak
Slovenian
Somali
Spanish
Swedish
Tajik
Thai
Tigrinya
Tonga
Turkish
Turkmen
Ukrainian
Uzbek
Vietnamese
Welsh
Wolof