How to learn AI and get RICH in the AI revolution

Learning to use AI tools like ChatGPT can 
make you more productive at your job. But,   learning to build AI tools like ChatGPT will 
make sure you have a job to be productive at. Will AI take over your job? Well, I don’t know 
about that. But one thing I do know is that   building AI tools would be one of the last jobs AI 
can replace. So, if you want to be future proof,   you might want to invest some time into 
learning to build AI. And if that’s not   a good enough reason for you, you will be 
shocked to know that OpenAI, the company   who built ChatGPT pays almost 1 million dollars 
to its AI engineers. In this video, I will give   you a step by step guide on everything you need to 
learn to be able to create AI tools like ChatGPT.   This video is especially important for someone 
who already knows a little bit of programming   or Math and wants to transition into 
an AI related job.

Let’s do this. Human Intelligence comes from the transmission 
of information through Neurons. Neurons are   nothing but a bunch of interconnected 
nodes in our brain. In the same way,   Artificial Intelligence also comes from 
a network of interconnected nodes called   Artificial Neurons. And these networks are 
also called Artificial Neural Networks or   simply Neural Networks. To be able to build AI 
tools like ChatGPT, we need to learn how to build   these Neural Networks. But to get there, 
we will have to take many smaller steps. You see, neural networks are part of this 
field called Deep Learning. In Deep Learning,   we take a neural network, train it by showing 
a lot of data and use the trained network to   predict something.

What to predict can vary 
depending on the task you assigned to the   network when you trained it. In the case of 
ChatGPT, the network is actually predicting   the next word of a sentence. I know it doesn’t 
make sense right now, but stay with me until the   end and I’ll explain this in much more detail. 
Moving on, Deep Learning is a subset of this   field called Machine Learning in which machines 
acquire the ability to learn. In other words,   Deep Learning is not the only way a machine 
can learn to predict. There are many other   ways to do that and all of them combined 
are part of Machine Learning. To be able   to learn Neural Networks and Deep Learning, 
we need to master Machine Learning first.

But how do we do that? Well, Machine 
Learning has 3 pillars. Mathematics,   Statistics and Programming. 
Let’s tackle them one by one. Let’s start with Mathematics because you will need 
good knowledge of Math to learn Statistics. Linear   Algebra, Calculus and Probability theory sit 
at the core of Machine Learning. And most of us   have been burnt by these either in high school 
or college.

But the good news is that you need   these concepts only to learn how different 
Machine Learning algorithms work. Once you   have learnt that, you will just write programs 
that implement these algorithms for you. So,   I don’t believe that you have to be a Math 
genius to be able to do Machine Learning. To   learn Math for Machine Learning, I recommend 
this specialization called Mathematics for   Machine Learning and Data Science on Coursera. 
This course is created by DeepLearning.ai which   was founded by Dr. Andrew Ng who is arguably the 
most well known professor of Machine Learning.   Later in the video, we will go back to Prof. Ng 
when we want to study Machine Learning. Anyway,   this specialization consists of 3 courses. One for 
Linear Algebra, another for Calculus and the last   one for Probability. If you think all of this is 
too overwhelming for you, you can also check out   this Data Science Math Skills by Duke University. 
This course is not as comprehensive as my other   recommendation.

If you are someone who doesn’t 
want to get too involved with the Math behind   Machine Learning or already knows the Math and 
just wants to brush up, this course is for you. Now that you know Math, let’s move on to 
Statistics. Now, Statistics is a very vast   field and it requires many many years to 
fully understand it. But luckily for us,   we don’t need to know everything for 
Machine Learning. For Statistics,   we will use a Breadth First Approach for learning. 
We will learn some basic core concepts and then   build upon them as we encounter new Machine 
Learning algorithms later. To learn all the   key concepts that you actually need, I recommend 
this course called Introduction to Statistics by   Stanford University. This course covers all the 
important ideas like Probability Distributions,   Central Limit Theorem, Confidence Intervals 
and Regression etc.

By the end of this course,   you would know all the Statistics you 
need to get started with Machine Learning. Before we can finally do some Machine Learning, 
we need to learn some programming. To be more   specific, we need to learn programming in Python 
because it’s the most popular choice when it comes   to Machine Learning. Now, there is some talk 
in the town about this new programming language   called Mojo which is compatible with Python and 
is 35,000 times faster. But it’s still too early   to predict the future of Mojo. So, we are going 
to stick with the time tested Python for now.

For   the purpose of Machine Learning, we don’t need 
advanced level programming skills. If you know   the basics like if statements, loops, functions 
and classes etc., you should be able to pick   Machine Learning easily. So, we are not going to 
build any crazy projects in Python at this step   and would focus on the basics. But we will build 
some Machine Learning projects using Python later   in the video. To learn these basics, simply go to 
learnpython.org and do some hands-on exercises. At last, we have reached a place where we can 
start doing some Machine Learning. We are just   one step away from building tools like ChatGPT 
now. For Machine learning, we need to go back to   Prof. Andrew Ng. Check out his Machine Learning 
Specialization on Coursera. This specialization   is divided into 3 courses. One for supervised 
learning algorithms like Linear and Logistic   regression.

Another one for unsupervised 
learning algorithms like Clustering. And   the last one for advanced algorithms that 
also introduces you to Neural Networks. If   you really want to have a deep understanding 
of ML, this is the best course out there. The   only caveat I would like to mention here is that 
the code samples and the Jupyter notebooks that   let you actually play with the code are 
not available for free with this course. Once you are done with the course, head over to 
Kaggle and do some hands-on practice. On Kaggle   you can see the projects that other people have 
built.

You follow along in the beginning and build   some confidence. When you are comfortable, you 
can participate in one of their competitions. This   will do 2 things. One, It will give you confidence 
that you can complete Data Science projects   independently. Two, you will build a portfolio 
of projects that you can write in your resume. Now that you feel confident about your Machine 
Learning skills, let’s go back to our original   goal which was to build AI tools like 
ChatGPT using Neural Networks and Deep   Learning. Machine Learning specialization by Dr. 
Ng already gives you an introduction to Neural   networks. But it’s not comprehensive enough 
for you to be able to understand advanced AI   systems like ChatGPT. For that, you will have 
to develop your skills in Deep Learning. But   there’s no need to worry because Dr. Ng also 
offers a specialization in Deep Learning.   First 3 courses in this specialization cover 
basics of Deep Learning which is basically how   to train Neural Networks.

In the fourth 
course, you will learn about Convolutional   Neural Networks which is basically Computer 
Vision. In Computer Vision, you will learn   how to train machines to recognize patterns 
in images which has applications in Autonomous   Driving and Face Recognition etc. But if your 
main goal is to understand how ChatGPT works,   that’s part of Natural Language Processing which 
is the last course in this specialization. This   course covers Transformer architecture 
which is what Chat GPT uses. By the end of   this specialization, you will know everything 
you need to have a successful career in AI. I know that this path can seem very long to many 
people. But, that’s the cost you pay to work on   the next generation technology.

Another path 
that is closely related to Machine Learning is   that of Data Science. In Data Science, you use 
data to develop insights but you don’t need to   be a Machine Learning expert. If you want to 
know the fastest way to learn Data Science,   watch this video. My name is Sahil 
and I will see you in the next one..