1
00:00:00,000 --> 00:00:09,120
[MUSIC]

2
00:00:09,120 --> 00:00:10,419
Welcome to the world of AI,

3
00:00:10,419 --> 00:00:13,920
get ready to peek behind the magic
curtain powering your favorite apps and

4
00:00:13,920 --> 00:00:18,450
gadgets through this introduction to
natural language processing or NLP.

5
00:00:18,450 --> 00:00:22,944
Have you ever wondered how Siri knows the
fastest route to the movies when you ask

6
00:00:22,944 --> 00:00:26,972
for directions or how Netflix seems
to know exactly which shows you'll

7
00:00:26,972 --> 00:00:29,890
binge watch next, the secret is NLP.

8
00:00:29,890 --> 00:00:32,535
It's the technology that
allows these apps and

9
00:00:32,535 --> 00:00:36,305
services to understand human language,
whether typed or spoken,

10
00:00:36,305 --> 00:00:40,290
without NLP you'd have to speak
to your devices like a robot.

11
00:00:40,290 --> 00:00:42,216
Let's take Netflix as an example,

12
00:00:42,216 --> 00:00:46,324
when you watch a Korean thriller one
weekend then a sci-fi series the next,

13
00:00:46,324 --> 00:00:51,380
Netflix uses NLP to analyze your viewing
history and make smart recommendations.

14
00:00:51,380 --> 00:00:55,344
It's like your own personal TV guide who
knows your tastes better than you do, or

15
00:00:55,344 --> 00:00:58,280
think about autocorrect on your phone.

16
00:00:58,280 --> 00:01:02,060
We all know how annoying it is when it
replaces a correctly spelled word with

17
00:01:02,060 --> 00:01:04,250
something totally random.

18
00:01:04,250 --> 00:01:08,892
But thanks to NLP, phones can now
understand slang, typos and abbreviations

19
00:01:08,892 --> 00:01:12,980
to fix your texting mistakes,
no more ducking auto-carrot fails.

20
00:01:14,110 --> 00:01:17,120
Ready to uncover more
real world NLP magic?

21
00:01:17,120 --> 00:01:21,442
Let's dive in and demystify how this game
changing technology understands human

22
00:01:21,442 --> 00:01:25,020
language to power the apps and
services we use every day.

23
00:01:25,020 --> 00:01:28,296
First things first,
what is natural language processing,

24
00:01:28,296 --> 00:01:31,638
natural language processing
abbreviated as NLP is a subset of

25
00:01:31,638 --> 00:01:36,760
artificial intelligence that intersects
with computer science and linguistics.

26
00:01:36,760 --> 00:01:40,231
Its primary objective is to enable
computers to understand, interpret,

27
00:01:40,231 --> 00:01:44,090
generate, and respond to human language
in a valuable and meaningful way.

28
00:01:44,090 --> 00:01:48,378
This understanding could range from simple
tasks such as identifying the language

29
00:01:48,378 --> 00:01:51,502
of the text to complex ones
like understanding sentiments,

30
00:01:51,502 --> 00:01:55,690
translating languages, and
even engaging in human like conversations.

31
00:01:55,690 --> 00:01:59,424
NLP encompasses a variety of
techniques and methods to analyze and

32
00:01:59,424 --> 00:02:03,160
represent natural language at
different levels of abstraction,

33
00:02:03,160 --> 00:02:08,310
from morphological and syntactic analysis
to semantic and discourse analysis.

34
00:02:08,310 --> 00:02:12,251
By processing and analyzing large
amounts of natural language data,

35
00:02:12,251 --> 00:02:16,258
NLP aims to extract information and
knowledge, or derive patterns and

36
00:02:16,258 --> 00:02:20,590
insights in a way that is similar
to how humans understand language.

37
00:02:20,590 --> 00:02:23,217
The ultimate goal of NLP
is to design algorithms and

38
00:02:23,217 --> 00:02:27,375
build systems that allow computers to
perform natural language related tasks,

39
00:02:27,375 --> 00:02:31,830
thereby bridging the communication
gap between humans and machines.

40
00:02:31,830 --> 00:02:36,517
Through the advancements in NLP, machines
can now assist in performing a plethora of

41
00:02:36,517 --> 00:02:40,097
tasks including but
not limited to automated customer service,

42
00:02:40,097 --> 00:02:44,920
sentiment analysis, language translation,
and content recommendation.

43
00:02:44,920 --> 00:02:49,105
Now that we've explored what natural
language processing is and its impressive

44
00:02:49,105 --> 00:02:54,340
capabilities in today's world, let's step
back in time to see where it all began.

45
00:02:54,340 --> 00:02:58,822
The journey of NLP is not just a tale
of technological advancement, but

46
00:02:58,822 --> 00:03:01,823
also a story of human curiosity and
ingenuity.

47
00:03:01,823 --> 00:03:06,167
From its earliest days of simple text
translation to today's sophisticated

48
00:03:06,167 --> 00:03:08,439
chatbots and complex language models,

49
00:03:08,439 --> 00:03:12,470
the evolution of NLP is as
captivating as its current state.

50
00:03:12,470 --> 00:03:16,496
So let's embark on a historical adventure
to uncover the roots of NLP and

51
00:03:16,496 --> 00:03:18,970
trace its path through the decades.

52
00:03:18,970 --> 00:03:21,571
Imagine if the conversations
we have today with Siri or

53
00:03:21,571 --> 00:03:24,962
Alexa were happening back when rock and
roll first hit the radio waves,

54
00:03:24,962 --> 00:03:28,850
that's how far back the story of
natural language processing starts.

55
00:03:28,850 --> 00:03:33,591
In the 1950s, the first steps towards
machines understanding human language were

56
00:03:33,591 --> 00:03:36,090
taken with the Georgetown-IBM experiment,

57
00:03:36,090 --> 00:03:40,510
which made headlines for translating
sentences from Russian to English.

58
00:03:40,510 --> 00:03:43,876
Around the same time the famous
Alan Turing proposed a test,

59
00:03:43,876 --> 00:03:48,366
now known as the Turing Test to see if a
machine could be considered intelligent by

60
00:03:48,366 --> 00:03:51,550
having conversations
indistinguishable from humans.

61
00:03:51,550 --> 00:03:54,071
As the 1960s rolled in
linguist Noam Chomsky's

62
00:03:54,071 --> 00:03:57,049
ideas helped shape how computers
dealt with human language,

63
00:03:57,049 --> 00:04:01,370
even though the actual language turned
out to be quite a puzzle for machines.

64
00:04:01,370 --> 00:04:04,744
Yet we saw programs like Eliza
in the 1960s that could mimic

65
00:04:04,744 --> 00:04:08,850
a therapist in a conversation showing
a glimpse of what was possible.

66
00:04:08,850 --> 00:04:12,578
The following decades were all about
building rules for computers to understand

67
00:04:12,578 --> 00:04:16,170
language and then teaching them
to learn these rules themselves.

68
00:04:16,170 --> 00:04:20,922
By the 1980s with machine learning coming
into play computers started getting better

69
00:04:20,922 --> 00:04:24,119
at understanding spoken words and
translating languages.

70
00:04:24,119 --> 00:04:28,502
The 1990s refine these methods with
computers getting better at figuring out

71
00:04:28,502 --> 00:04:30,417
the role of each word in a sentence.

72
00:04:30,417 --> 00:04:33,340
But it was the internet
explosion in the 2000s,

73
00:04:33,340 --> 00:04:37,992
that really gave NLP a playground of data
to learn from, leading to tools that could

74
00:04:37,992 --> 00:04:42,770
tell if a movie review was positive or
dig out information from heaps of text.

75
00:04:42,770 --> 00:04:46,911
The 2010s were a game changer with the
advent of deep learning, which allowed for

76
00:04:46,911 --> 00:04:51,370
even more advanced understanding and
generation of language by machines.

77
00:04:51,370 --> 00:04:52,300
Models like BERT and

78
00:04:52,300 --> 00:04:56,070
GPT showed us that computers could get
a lot better at handling language.

79
00:04:56,070 --> 00:05:00,590
Today, in the 2020s, NLP is not just
about technology it's also about thinking

80
00:05:00,590 --> 00:05:03,705
through the responsibilities
that come with it.

81
00:05:03,705 --> 00:05:07,980
Thanks to large language models,
chatbots like ChatGPT by OpenAI, Clawed

82
00:05:07,980 --> 00:05:12,775
by Anthropic, and Barred by Google are
helping improve how we talk to machines.

83
00:05:12,775 --> 00:05:16,719
They're bringing us closer to a time when
chatting with a computer will be as easy

84
00:05:16,719 --> 00:05:18,660
as chatting with a friend.

85
00:05:18,660 --> 00:05:21,850
Did you enjoy learning
about the history of NLP?

86
00:05:21,850 --> 00:05:25,564
Join me in the next video where I will
break down the building blocks that make

87
00:05:25,564 --> 00:05:28,310
up natural language processing,
I'll see you there.