How I'd Learn AI from Scratch (If I Could Start Over)
Here's how to learn AI in 2023.
If you want to dive deeper into AI, you should consider learning the technical part behind the AI tools that have become part of our everyday life.
Learning technical stuff for AI isn’t as simple as learning how to use ChatGPT though. You’d need to learn coding, math, machine learning, and other skills.
But don’t worry! You don’t need to be a genius in those areas to learn AI. In fact, you can start your journey today even if you don’t have any of those skills yet. In this article, I’ll share with you the path and resources that will help you learn AI from scratch.
Before we start, note that it’s not necessary to learn the skills below in any particular order. Explore each of them and find out which one you need to focus more on as you go.
Learn Key Python Libraries
The go-to programming language for data science and machine learning (ML) is Python. Although data science and ML are not the same as AI, they have some things in common.
It’s no surprise why Python is essential if you want to learn AI.
Now, Python is a language that is used in different areas, but, don’t worry, you don’t need to be a Python expert to learn AI but learn the basics and master some key libraries.
The approach to learning Python for AI might change based on your current coding skills.
If you have knowledge of other programming languages, I recommend you quickly review the basics of Python and then learn libraries such as Numpy, Pandas, Matplotlib/Seaborn, and scikit-learn. Why? Well, AI and ML apps work with data. By learning the key Python libraries mentioned before, you’ll be able to clean data, manipulate data, and make visualizations. That’s essential if you want to dive deeper into the world of AI.
There are many free options to learn Python and these key libraries. My favorite choice is the freecodecamp YouTube channel.
On the other hand, if you’re new to coding, before learning Python with tutorials, I recommend you use Brilliant to get started with programming and data analysis. I like Brilliant because it has interactive lessons to learn technical things from your phone. Highly recommended if you have zero knowledge of coding or data analysis.
Solve ML Projects
The best way to learn something is by doing. This means you have to work on projects to actually learn AI.
The good news is that there are a good number of projects available on the internet, but the problem is that most of them don’t have solutions, so you might get stuck in the middle of the project. That’s why I recommend you check out Kaggle projects.
There are a lot of ML project challenges in the Kaggle competition section.
The best part is that you have datasets available for the projects (collecting data would take you a lot of time in real life …) and you can check out other people’s code! Their solution is nicely presented in a notebook format, which means that the solution isn’t only code, but a step-by-step explanation of the approach they’re taking to solve the project.
If you’re learning AI alone, Kaggle will be your best companion when solving projects.
By the way, Kaggle also has a 3-hour Intro to Machine Learning course that you can access for free. That said, the course is a combination of articles and notebooks. If you’re more into video courses, I’ll share with you a free course to learn machine learning in the following sections.
Learn The Necessary Math
Math is a skill we need to acquire to understand the algorithms and more stuff that’s behind ML and AI. That said, we don’t need to become expert mathematicians for this but to have high-school-level math.
I said high-school-level math, so you don’t go crazy and learn too much math because, in the end, you might need some concepts more than others. That’s why it’s a good idea to stick with the basics in the beginning and learn more math as you go.
I believe most of you have (or had) that high-school-level math needed for machine learning, but in case you haven’t solved a math problem for a long time and want to review the stuff you’ll need, here are some resources.
Khan Academy: They have a bunch of free video lectures about calculus, statistics, probabilities, etc.
Brilliant: They have interactive lessons to learn the high-school math you need for machine learning.
I’d recommend Khan Academy for those who need to review the math concepts needed for ML from scratch, and Brilliant for those looking to brush up on their math skills.
Free Machine Learning Specialization
To dive deeper into the world of ML and AI, I recommend you take this Machine Learning Specialization on Coursera. It’s completely free (unless you want to get a certificate) and it’s instructed by Andrew NG, a recognized leader in ML and AI.
This is a beginner-level specialization that contains a 3-course series: supervised machine learning, advanced learning algorithms, and unsupervised learning.
In my experience, you need to have a solid knowledge of coding and math before taking this course. Overall, in the course, you’ll learn to build ML models with the key Python libraries I mentioned before, use supervised and unsupervised learning techniques, build neural networks with TensorFlow, build recommender systems, and more!
According to Coursera, if you dedicate 10 hours a week to this course, you can finish it in 2 months, so if you’re really into ML and AI, start today!
The sky is the limit!
As you dive into the world of ML and AI, you’ll realize that there’s something new to learn to develop further in a particular area. It can be a new library, new ways to efficiently collect data, good practices in data cleaning, new math concepts, new algorithms, etc.
That’s a good sign that you’re making progress in your journey, so don’t be afraid to keep learning new things as you go.
The following are not artificial intelligence: machine learning, deep learning, LLMs. There are other attempts such as CYC in Austin, Texas. Doug lenat, the developer just died so there may be many articles on his work. Lenat and Marcus: https://arxiv.org/abs/2308.04445