1
00:00:01,170 --> 00:00:04,730
- IBM refers to its
Watson capabilities as a

2
00:00:04,730 --> 00:00:08,270
cloud-based, cognitive computing platform.

3
00:00:08,270 --> 00:00:10,760
And basically what they mean by that is,

4
00:00:10,760 --> 00:00:14,670
their systems try to
learn and make decisions

5
00:00:14,670 --> 00:00:17,070
the way the human brain works.

6
00:00:17,070 --> 00:00:19,800
They consume massive amounts of data

7
00:00:19,800 --> 00:00:23,460
and then train their
systems based on that data

8
00:00:23,460 --> 00:00:27,370
to try to make decisions, and
they're deploying this now

9
00:00:27,370 --> 00:00:32,120
across a massive range of
real-world applications.

10
00:00:32,120 --> 00:00:33,620
Now, there are a couple of

11
00:00:33,620 --> 00:00:37,221
artificial intelligence
milestones that IBM had,

12
00:00:37,221 --> 00:00:40,810
that basically captured
people's attention,

13
00:00:40,810 --> 00:00:44,640
made them start thinking
about intelligent computers,

14
00:00:44,640 --> 00:00:48,270
and also started to make
businesses think about

15
00:00:48,270 --> 00:00:52,430
commercializing artificial
intelligence capabilities.

16
00:00:52,430 --> 00:00:56,750
Way back in 1997, a predecessor to Watson,

17
00:00:56,750 --> 00:01:00,010
IBM's DeepBlue computer beat the reigning

18
00:01:00,010 --> 00:01:02,740
world chess champion, Gary Kasparov,

19
00:01:02,740 --> 00:01:05,140
and that was done under
tournament conditions.

20
00:01:05,140 --> 00:01:09,410
And then back in 2011,
Watson playing against

21
00:01:09,410 --> 00:01:12,530
the two best human Jeopardy!
players in the world,

22
00:01:12,530 --> 00:01:15,910
won a $1 million Jeopardy! match.

23
00:01:15,910 --> 00:01:20,910
And interestingly, they trained
the Watson computer for that

24
00:01:21,640 --> 00:01:24,870
using a variety of techniques
from machine-learning,

25
00:01:24,870 --> 00:01:27,240
which we'll look at in a subsequent part

26
00:01:27,240 --> 00:01:29,650
of this presentation, and also

27
00:01:29,650 --> 00:01:31,840
reinforcement learning techniques,

28
00:01:31,840 --> 00:01:34,920
which are techniques
that are very similar to,

29
00:01:34,920 --> 00:01:37,423
for example, how we as humans learn.

30
00:01:39,174 --> 00:01:41,982
For instance, as a baby when
we're first learning to walk,

31
00:01:41,982 --> 00:01:46,180
and we kind of get up and then
we're holding on to something

32
00:01:46,180 --> 00:01:48,440
and then we let go and we fall down,

33
00:01:48,440 --> 00:01:51,560
that is something that
registers in our brain,

34
00:01:51,560 --> 00:01:55,360
and then we get up the next
time and maybe we hold on longer

35
00:01:55,360 --> 00:01:58,760
and so we reinforce the fact
that if we're holding on,

36
00:01:58,760 --> 00:02:01,450
it helps us stand up, and eventually

37
00:02:01,450 --> 00:02:05,180
we learn to start walking
and we may find that

38
00:02:05,180 --> 00:02:08,760
as we fall to one side there
is something that we need to do

39
00:02:08,760 --> 00:02:11,780
to re-balance ourselves,
to go back the other way,

40
00:02:11,780 --> 00:02:14,940
and as we make mistakes, we
learn from those mistakes

41
00:02:14,940 --> 00:02:18,790
and we keep reinforcing
our learning process,

42
00:02:18,790 --> 00:02:21,570
and that technique was also used

43
00:02:21,570 --> 00:02:24,670
as part of training IBM Watson.

44
00:02:24,670 --> 00:02:27,130
Now, they also used a variety of

45
00:02:27,130 --> 00:02:28,970
language-analysis techniques,

46
00:02:28,970 --> 00:02:31,750
hundreds of different
language-analysis techniques,

47
00:02:31,750 --> 00:02:36,100
and they loaded up 200
million pages of content,

48
00:02:36,100 --> 00:02:39,220
including all of Wikipedia at that time

49
00:02:39,220 --> 00:02:41,485
for the purpose of training Watson

50
00:02:41,485 --> 00:02:45,137
to compete in the context of Jeopardy!.

51
00:02:46,050 --> 00:02:50,360
Now, one of the reasons we
chose to feature IBM Watson

52
00:02:50,360 --> 00:02:54,340
in this set of videos and
our corresponding books

53
00:02:54,340 --> 00:02:58,670
is that they have this
no-credit-card-required policy

54
00:02:58,670 --> 00:03:02,090
so that you can sign up
for an account and access

55
00:03:02,090 --> 00:03:04,530
their free Lite tier services,

56
00:03:04,530 --> 00:03:07,020
which makes it really convenient and easy

57
00:03:07,020 --> 00:03:10,120
for both students and
professionals to experiment

58
00:03:10,120 --> 00:03:13,680
with these incredibly
interesting capabilities.

59
00:03:13,680 --> 00:03:15,970
As you'll see over the next many videos,

60
00:03:15,970 --> 00:03:19,490
Watson provides a lot
of interesting things,

61
00:03:19,490 --> 00:03:23,640
intriguing capabilities,
such as being able to perform

62
00:03:23,640 --> 00:03:26,340
natural language translation,
one of the features

63
00:03:26,340 --> 00:03:28,760
that we'll be using later in this lesson,

64
00:03:28,760 --> 00:03:32,250
converting speech into text, and similarly

65
00:03:32,250 --> 00:03:34,300
converting text into speech.

66
00:03:34,300 --> 00:03:39,150
They have web services that can
understand natural language,

67
00:03:39,150 --> 00:03:41,860
web services that can help
you implement chatbots,

68
00:03:41,860 --> 00:03:45,560
so for instance, if you're
responsible for customer service,

69
00:03:45,560 --> 00:03:49,750
the first line of interaction
nowadays through a website

70
00:03:49,750 --> 00:03:53,630
is often a chatbot that
helps route the person to the

71
00:03:53,630 --> 00:03:56,950
correct customer service department.

72
00:03:56,950 --> 00:03:59,410
There are capabilities for looking at text

73
00:03:59,410 --> 00:04:03,360
and analyzing it for tone,
and other capabilities

74
00:04:03,360 --> 00:04:06,060
that allow you to look
at both static images

75
00:04:06,060 --> 00:04:10,030
and full-blown video, and recognize things

76
00:04:10,030 --> 00:04:12,858
like objects within
those images and video,

77
00:04:12,858 --> 00:04:15,280
colors, people, et cetera.

78
00:04:15,280 --> 00:04:19,570
And that's just a tiny subset
of the capabilities available

79
00:04:19,570 --> 00:04:24,380
through Watson and then the
greater IBM Cloud as well.

80
00:04:24,380 --> 00:04:28,100
Now, in this lesson we're
going to take a look

81
00:04:28,100 --> 00:04:30,610
at setting up an IBM Cloud account.

82
00:04:30,610 --> 00:04:33,800
I will demonstrate a
few of the Watson demos

83
00:04:33,800 --> 00:04:36,508
that are provided by IBM directly.

84
00:04:36,508 --> 00:04:39,880
In particular, we will
be taking advantage of

85
00:04:39,880 --> 00:04:44,300
three of the Watson services
later on in this lesson.

86
00:04:44,300 --> 00:04:47,440
We'll show you how to install
the Watson developer cloud,

87
00:04:47,440 --> 00:04:50,860
Python SDK, which is how we'll
programmatically interact

88
00:04:50,860 --> 00:04:53,450
with the Watson services,
and as you'll see,

89
00:04:53,450 --> 00:04:57,050
we're going to develop
this traveler's companion

90
00:04:57,050 --> 00:05:00,180
translation app, which is
going to mash up several

91
00:05:00,180 --> 00:05:02,554
Watson web services, specifically

92
00:05:02,554 --> 00:05:07,270
their translation service,
their speech-to-text service,

93
00:05:07,270 --> 00:05:09,470
and their text-to-speech service,

94
00:05:09,470 --> 00:05:12,520
and we're going to create
an app that enables people

95
00:05:12,520 --> 00:05:15,490
who only speak English
and only speak Spanish

96
00:05:15,490 --> 00:05:17,860
to communicate with each other directly,

97
00:05:17,860 --> 00:05:20,140
despite that language barrier.

98
00:05:20,140 --> 00:05:22,850
And as you'll see, it's
relatively straightforward

99
00:05:22,850 --> 00:05:24,960
to take advantage of these services

100
00:05:24,960 --> 00:05:28,480
because of the powerful
classes provided by

101
00:05:28,480 --> 00:05:32,223
the Watson developer, Python SDK.