How I'd Learn AI (If I Had to Start Over)

if you want to jump into artificial intelligence and build things with AI today how should you start things are so fast changing my stomach hurts whenever I think about it loads of open- source tools and models are being created every single day Microsoft co-pilot is being launched very soon which brings AI to all Microsoft Office applications so if your hands touch keyboard for work this is going to change your job in the next few years this motivates me if there's something that's going to change my life and work forever I want to learn everything about

it I want to learn how AI is built how to build with AI and how to use it to enhance my work I think as someone working in the tech field you want to be in the position where you can build things with AI rather than just consume it so today I'll be sharing with you a road map as a One-Stop shop for you to expand your AI skill set learn the fundamentals of AI and learn to build with AI even if you don't have any background in machine learning mathematics statistics or programming I hope

you walk away with some useful ideas of where to start let's get started so why should we bother learning AI this whole AI machine learning and deep learning thing has been around since the 1950s the AI we often talk about today is generative AI which is a subset of machine learning and deep learning generative AI can now write codes generate stunning images writing music diagnos rare conditions creating outline for presentations reading images and much much more companies are looking into implementing AI solutions to solve their specific problems this is a gold mine because everything is

still so new so if you have the knowledge and know how to build things with AI you can create huge impact as with anything in its early days AI model still have many issues that need to be solved they are not yet reliable or stable they potentially possess biases among other things that's why we need more people who have the in-depth understanding and can get to the bottom of the Technologies to solve various problems even if you don't build things yourself the knowledge can help you avoid a lot of misunderstanding and misinformation such as AI

can do everything as long as you got the right plugins okay when it comes to learning AI of course there's no one- siiz fit all road map you might have seen a lot of low codes or no code tools to help you get started with AI and even develop things like an AI chatbot you can certainly play with them to have the first feeling of how things work and what is possible and even build great real world Solutions with them as you get more into it you might realize that those tools might feel like a

little bit of a black box they might not be so flexible that you can customize a lot of things or sometimes your solution works and sometimes it fails so if you really want to learn how artificial intelligence works and build tools that are reliable scalable and can be tailored to your specific use case I think that's when you hit the ceiling and can't rely solely on those low code platforms anymore personally I often prefer to have a solid understanding of how things work from ground up so let me walk you through a road map for

learning AI you can download this full road map in PDF together with my recommended learning resources you can find the link in the description below on the fundamental level you want to learn the basics of programming python is one of if not the most used programming languages for machine learning deep learning and AI so some coding knowledge in Python would be essential for working with python you can certainly use several Ides or integrated development environments for example Visual Studio code by charm or Jupiter notebook Jupiter notebook is my favorite environment to start learning python here's

a simple setup of a python project in Jupiter notebook and here's the same project in Visual Studio code you can use any of these tools and they are completely free if you've never coded in Python before make sure to at least get yourself familiar with the four basics of python firstly data types and the operations that you can do on them data structures and how to work with them conditionals loops and functions and lastly objectoriented programming and using external libraries when learning python the most essential libraries to learn are numai for computing and working with

numerical data pandas for wrangling table data or data frames when you're already familiar with those libraries you can start learning some other libraries that you can use for your AI project for example M plot lip is a popular library for data visualization Spacey for basic text processing functionalities and so on for working with large language models Lang chain is a very useful library to learn to develop multiple applications on top of airms what's nice with python is that there are so many open-source libraries that you can use to develop almost anything you want the next

thing I'd recommend you learning is Git Version Control if you're not yet familiar with it git is an open-source software for tracking changes in your project basically managing versions of your project Version Control is essential when you're collaborating with other people in a large or complex project they're actually just a few Concepts you need to understand to start using git as shown in this diagram the funny thing is that many people actually confuse git with GitHub GitHub is a hosting platform for git repositories so that you can share your project with other people across the

internet meanwhile git is the software itself through GitHub you can directly see and contribute to other people's project so how can you start using git the easiest way is to install GitHub desktop stop is a user interface tool to help you work with Git or if you prefer to use the terminal you can also interact with Git through the terminal comments I always keep a small cheat sheet here to remind me of some common G comments which I find super convenient and helpful another essential thing to learn in your AI journey is using apis knowing

how to use apis is a magical skill that opens up a whole new world of possibilities API stands for application programming interface it's a way for computer programs to communicate with each other basically there are two Jons you need to learn yeah just two API request which is also referred to as API call and API response depending on the API you can make a request for data or for model prediction in the case of chat CPT without knowing how to use the API you'll be limited to the chat interface on the open a website don't

get me wrong I love the chat to be website but you can't develop your own tool this way or integrate the AI model into your current system okay on to the next level we are concerned with the theoretical fundamentals of AI I recommend you at least get some highlevel theoretical understanding of AI and its subfields such as machine learning neuron networks and deep learning and optionally computer vision n and reinforcement learning okay if we look at this diagram again deep learning is a subset of machine learning traditional machine learning algorithms mostly fall into either supervised

learning this is when you actually have the target labels to train the prediction model on or unsupervised learning when there's no target labels in general these algorithms only work for Tabo data think of data tables with each record Being a row and each data feature being a column in that table if you're interested you can quickly go over some machine learning jargons and get some high level understanding of these algorithms they're all cool to learn but since AI today uses deep learning I think to save time you can probably just jump right into deep learning

I think you will anyway learn the essential machine learning Concepts along the way as we just briefly talked about neuron network is the algorithm behind deep learning it works incredibly well for unstructured data like text and images for neuron Network you want to understand the main Concepts such as forward propagation back propagation gradient descend algorithm and how weights are updated in the network if you like some math you can try to understand all the calculations underlying the neuron Network they are really not too difficult to understand so don't shy away from it neuron Network in

itself is pretty simple and maybe even a bit inferior from the mathematical sta discal point of view no offense here to the fathers of deep learning but when you stack many many Network layers together into a complex architecture this is when things get interesting the neuron Network now can start recognizing digits classifying cats and dogs to predicting the next tokens in the case of LMS today convolutional neuron networks use to be a very popular architecture for deep learning with images because it can recognize patterns on images and record neuron networks used to be very popular

for text modeling because it can understand sequences however these architectures have become pretty upsolid since the invention of Transformers architecture in 2017 this is the architecture behind the foundation language models nowadays Transformers outperform pretty much all the earlier architectures so you might want to jump into it right away and reverse engineer the knowledge if you see any Gap as you work with the AI models you also want to get yourself a high level understanding of how the foundation models underlying them are trained in general they trained through an unsupervised process it's up to you to

learn more about the nitty-gritty details of training your language models when working with language models you might hear the common term text embeddings this is a very useful concept to understand text embedding converts text into vectors of numbers nowadays we often forget that computers cannot actually understand human languages computers can only understand numbers so this conversion step is really necessary there have been many embedding models created with ever SM ways to capture meanings into those vectors no matter what you're learning or where you are on this journey you can build relevant projects to get your

hands dirty and experiment with things this will help you connect the dots and challenge your own understanding for example if you're learning python you can build your first neuronetwork in Python using kasas or tensorflow library is only a few lines of code if it's too to high level you can try to write a neuron Network and Implement gradient descent from scratch with numai when you're learning the theories a real world project would be to pick one specific concept you find interesting and write a blog post or make a video on it this will help you

understand the concept deeper and help other people too if you're ready to tackle more complex AI project you can build a real world application for example you can create a document retrieval app basically to create a chat with PDF kind of application where people can upload a document and ask specific questions based on the documents or you can create your own chatbot I also want to mention that no matter what project you do big or small make sure you document them for reference for yourself and share that with other people through articles and social media

posts you really never know how many people might find it useful the next thing I'd recommend is to develop mental models around Ai and perhaps specialize in a certain area within AI if you want to I think reading books about AI is a great way to go through the noise on social media and get a more well-rounded background of AI this also equips you with the right Frameworks and tools to reason and interpret things that you see or hear about AI today you can also find my AI book list in the road map personally I

also find it so crazy how much important stuff around AI that is not talked about more widely on the mainstream media there are many topics around AI that don't make headlines for example Advanced prompt engineering methods to improve the quality of the airm response like self-consistency chain of thoughts prompting or automatic prompting autogen project by Microsoft that allows you to develop LM applications using multiple agents that can converse with each other to solve tasks Advanced document keyway with multimodel documents that can work well with complex table taes images and other data structures AI security and

hiking the other day I watched a YouTube video of researcher who uncovers some serious security issues with machine learning models this is a very overlooked area until now so if you know computer security very well please do Humanity your favor and look into this we also have ai Safety Research that is the area to find ways to align ai's goals with humans goals because if we fail to do this we are literally screwed and finally AI regulations if you're interested in laws in Europe the EU AI Act is one of the big things that are

coming to regulate the use of AI the US government also recently passed the executive order on the safe secure and trustworthy development and use of artificial intelligence to address the potential risks of AI no one knows how helpful these regulations are going to be so if you want to dive into any of these areas it's generally quite easy to find information on these Topics by reading books research papers articles and watching videos on my channel for example I also follow some newsletters like the bat newsletter by Andre in to get informed about some important AI

developments medium is also where I often find really useful articles I sometimes dive into research papers as well to learn about some new cool research my friend Sophia Yang has an awesome Channel where she breaks down all the cool research around AI so make sure to check out her Channel if you like this stuff we are still in very early days of AI and we don't know how things will turn out in the next few years but one thing we know for sure is that things are changing faster and faster the only way to keep

up is to continuously learn so I hope you found this video helpful and if you like the video smash the like button and subscribe to my channel for future content around data science and AI thank you for watching bye-bye [Music] I