< DY PATIL >

1 GENERAL INFORMATION ABOUT THE COURSE

1.1.1.1 Course Overview
This Introduction to Watson AI will introduce you to the IBM portfolio of Arti-
ficial Intelligence (AI) services, tools, and APIs.

1.1.1.2 Who Should Take This Course

This course is designed for everyone who wants to learn about Implement-
ing AI solutions with IBM Watson. You do not need programming experi-
ence for this course. Whether you are a developer, an executive, or a stu-
dent, this is the place to learn about IBM Watson.

1.1.1.3 Pre-requisites
The course assumes you have a basic understanding of AI and are familiar
with its terminology.

We recommend that you complete the following course before starting this
one:

 AI for Everyone: Master the Basics

1.1.1.4 Changelog
01 Sept 2020 (Rav Ahuja): Updated version of the course published on edX.org

01 Sept 2020 (Anamika Agarwal): Replaced links to labs with links from SN Asset Library

2 SYLLABUS
Syllabus

Module 1 - Watson AI Overview

 Let's Put Smart to Work

 Watson AI Professionals
 Ways Watson Can Help Your Business
 Quiz

Module 2 - Watson AI Services

  Introduction to Watson Services
 AI Assistant
 Knowledge
 Speech
 Language
 Quiz

Module 3 - More Watson AI Services

 Data
 Vision
 Empathy
 Compliance
 Quiz

Module 4 - Watson in Action

 Watson Use Cases

 Quiz

Final Assignment - Watson in Action

3 GRADING SCHEME
3.1.1.1 GRADING SCHEME
This section contains information for those earning a certificate. Those
auditing the course can skip this section and click next.

1. The course contains 5 Graded Quizzes, 1 per each module and 1 for
the final assignment. Each Graded Quiz carries an equal weight of
20% of the total grade.
2. The minimum passing mark for the course is 70%.
3. Permitted attempts are per question:
o One attempt - For True/False questions
 o Two attempts - For any question other than True/False
4. There are no penalties for incorrect attempts.
5. Clicking the "Final Check" button when it appears, means your sub-
mission is FINAL. You will NOT be able to resubmit your answer for
that question again.
6. Check your grades in the course at any time by clicking on the "Pro-
gress" tab.

4 MODULE INTRODUCTION
In this Module, you will learn how Watson AI works. You will understand the
many ways Watson AI is helping professionals and businesses reimagine
their workflows, learn more from less data, and protect their insights.

5 LEARNING OBJECTIVES
 Describe how Watson AI works.
 Describe ways in which Watson helps professionals.
 Describe 3 ways in which Watson helps businesses.
 Explain the difference between Consumer AI and Business AI.
 Explain the 3 steps to getting started on your AI journey.

6 WATSON - THE AI FOR BUSINESS

So the whole world is talking about AI. Big bold promises like.
It'll transform the human race. It's going to solve unsolvable problems.
It'll find life on Mars. But here's the thing. But you don't live on Mars. Today
you are building wind turbines. Good morning sir. See if the blade is
passing quality gauge. That's why you work with Watson. I detect frictional
loss on the mid span. It can detect the tiniest defects from just a few im-
ages to help production stay on time, and on budget. I optimize a fiberglass
finish to reduce frictional loss and maximize airflow. Welcome to the Ka-
roon gas plant, operated by Woodside. What you are seeing is
a world-class gas production facility located on the Northwestern coast of
Australia. Working in these challenging conditions, it is important that ques-
tions are answered in minutes not hours, especially for those working on
operational sites. That is where I come in. Natural language and conversa-
tion APIs, allows me to understand the many projects related
to questions which employees can ask via their tablet or computer. All they
have to do is ask. Using my retrieving rank API, I can find the answers by
reading through tens of thousands of project related documents in seconds.
And with Watson Explorer, I can apply expertise that I have learned from
thousands of employees. Working together with Woodside, I have helped
reduce time spent tracking down expert knowledge by 75 percent. Not bad
for the new guy. America's heartland, these crops provide our meals. But
the journey to the dinner table begins where you least expect it. It all starts
with a bee. Every third bite of food we eat is from a pollinated plant.
Life for beekeepers, farmers, even you and me, is about to change for the
better. IBM's Watson has joined forces with The Weather Company. That
means more precise forecasts, so people like Richard, can make better de-
cisions. Moisture, for example, is critical. If it gets too dry out in fields
the blossoms dry up and bees won't work. By learning from data provided
by sensors, Watson can help irrigation systems get water where it's needed
most. It's the same way some bees forage while others collect water.
But even if the weather is right, getting bees to the crops is a challenge.
Last year, Richard sent 150 semi loads of bees to California for pollination.
To keep the bees safe, the trucks can only travel during the day.
Like bees flying to flowers, Watson can help identify the best routes that
take traffic, temperature, and road conditions into account. Because that
could mean a smoother ride for bees. Watson isn't just the promise of AI, is
the real deal deep learning business AI, that knows your industry, protects
your insights, and works with tools that you already use.

7 HOW WATSON WORKS

AI is changing the way the world works, making business faster, smarter,
and more secure. At IBM, we're helping companies put AI to work at scale,
giving them an unparalleled business advantage. With Watson, businesses
are personalizing customer experiences, streamlining processes, minimiz-
ing risk, and sparking innovation. With Watson, engineers are keeping mil-
lions of elevators moving, predicting when they will break down, and pro-
actively fixing them. With Watson, banks are deploying virtual agents
trained on thousands of customer inquiries, helping them provide expert
service to millions of customers, 60 percent faster. With Watson, a range of
industries from health care to automotive, to telecom, to education, are
working faster and transforming workflows. Businesses everywhere are
building their future with Watson. Watson is an artificial intelligence system.
Artificial Intelligence is what gives machines the power to learn, adapt to
new inputs, and make better decisions. Machine Learning, a subset of AI,
uses computer algorithms to analyze data and make intelligent decisions
based on what it's learned. It's how streaming sites recommend new music
by comparing what one listener likes to others with similar tastes. Watson
uses a sophisticated machine learning technique called Deep Learning.
Deep Learning layers algorithms to create an artificial neural network. It
can continuously learn on the job, determining whether decisions were cor-
rect, constantly improving the quality and accuracy of results. This is what
enables it to learn from unstructured data such as photos, videos, and au-
dio files. Meaning Watson can tell an insurance company exactly how a
car's been damaged in a crash, by discerning the make and model before
comparing images of an undamaged car to the damaged one, even getting
as detailed as a broken exhaust. Deep Learning is also what enables Wat-
son's Natural Language Understanding capabilities, allowing it to learn by
deconstructing sentences,

then analyzing and identifying the concepts and relations of those sen-
tences. Once it understands how everything fits together, it can work out
the context and intent of what's being conveyed. But your AI also needs to
understand the specific language and terminology of your industry. With
traditional AI models, this would require lots of additional data and comput-
ing power, taking weeks or months. With Watson however, it can be done
quickly and simply, thanks to a technique called Transfer Learning. Trans-
fer Learning is learning how to learn. It's what enables Watson to learn
more from less. so it doesn't need to be trained from scratch. It can be fed
prior knowledge to speed things up. Let's take the case of a home insur-
ance company, to demonstrate how Watson's Transfer Learning is a three-
layered AI model. The bottom layer is made up of out of the box general
knowledge, like Wikipedia for AI. This layer tells Watson what a house is, or
a tornado. The middle layer is pre-packaged with knowledge tailored to
specific industries. So for insurance, it's how Watson understands terms
like coverage or beneficiaries. The top layer is where the customer specific
learning takes place, personalizing Watson to the company's unique busi-
ness needs. Fed with their data and know-how, the model can understand
a company's specific risk attributes, and behavioral policy pricing models.
Transfer Learning is a key part of how Watson is already accelerating
claims processing by 25 percent, dramatically cutting operating costs.
Transfer Learning makes Watson the best AI for your enterprise. It allows it
to learn faster from smaller sets of data, while also protecting your data and
insights. Watson is designed to ensure data only flows from the bottom to
the top. So you can make the most of the shared knowledge in the bottom
two layers, while retaining ownership of your top layer. Most companies AI
systems use your data to build their general machine learning models, al-
lowing all their users to benefit from that collective knowledge. Watson's
three-layer model ensures your data, remains your data.

8 ENRICHING CUSTOMER SERVICE

With Watson, businesses are personalizing customer experiences. There
are several different approaches clients can take to enrich their customer
service experience. One, is to have a 24/7 assistant to interact directly with
customers, while the other is having a behind the scenes assistant to help
employees and how they engage directly with their customers. IBM's Wat-
son Assistant, a virtual assistant powered by underlying natural language
models, understands your users and provides training recommendations as
you build and run your assistant. We picked Watson mainly because it was
the most extensive platform available and regardless of what we wanted to
do, there was some solution that IBM and Watson was able to deliver for
us. Ava is a custom application which we built on our side but the brains of
Ava has effectively Watson. Our vision is that Ava essentially is the first
stop for any customer interaction in the future, because we know that if
nothing else, hopefully Ava can understand what the customer is asking for
and even if we need to direct them to a person, one of our human agents,
at least we are starting from a place where they know what the customer
needs. Our resolution time with Ava is about five minutes. A resolution time
with our people is about 38 hours. On the cost side, it's between $15 and
$200 per case, with human agents, roughly a dollar when it comes to Ava
interactions. Is customer satisfaction, roughly the same between an Ava in-
teraction and a human interaction? Ava has already done over 30,000 con-
versations this month alone. The future of customer service will change and
I think that while it may seem a little bit odd to have an artificial agent talk-
ing to a customer now, three or four years from now it will seem very odd if
you don't have one. If we can help more customers than we ever could be-
fore, then you're delivering the value that you hope to by doing this. Banks
are deploying virtual assistants trained on thousands of customer inquiries,
helping them provide expert service to millions of customers, 60 percent
faster. Royal Bank of Scotland's home buying and ownership business has
employed CORA and MARGE, both powered by Watson to deliver better
customer service and to empower the context center agents. We wanted to
make it easy to get help without customers having to pick up the phone or
look for answers in too many places. CORA takes the routine stuff off our
plates and helps us be brilliant in the moments that matter most to our cus-
tomers. The great thing about, a virtual agent is, you can see all the con-
versations that she's having with customers. When developing CORA, we
analysed data to determine the most commonly asked questions. I don't re-
cognize this transaction? How do I pay my bills online? I want to change
my address. CORA's trained with over 200 customer intents and has over a
thousand responses to those intents. The simple questions. I'd like to
check my balance? CORA responds with the answer straight away. For
more challenging questions, we offer guidance to the customer so that they
can help themselves. And for really complicated questions: How do I make
my first mortgage payment? CORA seamlessly hands off the customer to
an agent. When customers get to me, I have that information already. I
think our AI journey has been fantastic, and it's going to allow us to create
new products and new benefits for our customers for years to come. Wat-
son Assistant automatically directs requests down the optimal path for solv-
ing the problem. It knows when to provide a direct answer to a common
question or reference more generalized search results for something more
complex. It also helps agents find answers to complex questions faster.
Watson Assistant is powered by underlying natural language models that
understand your users and provide training recommendations as you build
and run your assistant. The foundation of Watson Assistant addresses four
key principles. First, it comes pre-trained with industry-relevant content.
The second is that, it remembers and understands user input. For example,
your customer's having a conversation with Watson and Watson recog-
nizes a new intent, Watson needs to answer that question and remember
this conversation just as you would expect a human to remember. Third, it
recommends. Watson assesses what is happening in these conversations
and recommends places that may be strengthened by having other intents.
When you provide a few example of questions, intents, and clarifying
terms, entities. Watson Assistant can build a machine learning model to un-
derstand similar natural language requests from your users. The fourth
principle is about protecting your data. It ensures that your data is used to
train Watson for you and not for your competitors. With Watson, it is also
possible to integrate with customer care platforms, making it the best AI for
personalizing customer experience.
9 INTELLIGENT CHATBOTS, VIRTUAL ASSISTANTS, AND
IBM WATSON
Can you talk a little about chatbots? Sure thing. Now chatbots are one of
the most popular areas when it comes to machine learning technology. It's
one of the things that's gotten the most attention recently. I mean really to
drive across the point of why chatbots are so useful. Imagine this, you're on
a call with your ISP, your Internet Service Provider, because the Internet's
down, and you're greeted with this annoying robot voice that says, press
one if so-and-so, press two if so-and-so and there's tons of options that
you've got to listen through every single one, the options don't fit your
needs, you have to keep repeating until finally connected to a human and
you're not even sure if it will. That's an annoying scenario, and it's some-
thing that all of us have gone through at one point or another. But if you
think about it, companies themselves also are in kind of a bind here. They
can't hire over excessive numbers of people to work in their customer ser-
vice departments. But at the same time they need to keep it intuitive, they
need to keep it simple, they need to balance things, and this is where chat-
bots come in. With chatbots, you can balance the interaction required
between humans and bots. For example, Watson has special Natural Lan-
guage classification techniques that enable it to understand different kinds
of sarcasm, different puns, and so with chatbots, essentially what you're
doing is you are balancing out human and machine interaction. When you
call your ISP now, you're going to be greeted by an intelligent bot that can
understand what you need. If you want to be connected to a human in-
stantly, it will understand that request of yours and connect you to a human
as soon as possible. Or if you've got a concern that it can help you out with
directly, it will converse with you in a natural way, in an implicit way which is
how we as humans have thought since the beginning of human history, in
order to have this sort of intuitive method of solving your query. So with
chatbots, essentially what you're doing is you're saying we're not going to
replace humans outright, but we're not going to keep only humans. We're
going to make it so that the vast majority of simple questions can be filtered
through this chatbot, but a few of the complex questions that would require
creativity, or exceptions, or things that aren't part of the rules or things like
this, will then go to the humans so that they can actually focus on fulfilling
work. The people at the customer service representatives aren't being over-
worked, they're not being bored by answering the same questions 20 times
in a row, and at the same time people have a more intuitive experience with
the bots and the company gets to save money. It's a great solution, all in
one, it's a win-win situation for everyone involved. So chatbots are very
popular, they do deserve that hype, and I can't wait to see how people start
to use chatbots in the next few months to years.