Analytics Vidhya
Published in

Analytics Vidhya

How to Break Into AI? Harvard CS50AI: Introduction to Artificial Intelligence Course Review

A notebook saying “notes”
By David Travis on Unsplash

If you are reading this, you are probably looking for the door into the field of Artificial Intelligence. I am glad to say that you have found it. CS50AI, a part of the Professional Certificate program on edX in pair with CS50, is a great source that takes you along the spectrum of available AI topics.

I recommend the course to anyone looking to dip their toes in the field and see if that is something you would enjoy. The course starts with basic algorithms, such as Breadth-First Search, and ends with comprehensive concepts like Machine Learning and Deep Learning.

CS50AI Certificate

Let’s answer some basic questions before we begin.

How long? 7 Weeks (One lecture each week)

How much effort? 10–30 hours per week (30 hours is more like it)

How much does it cost? $200 for a verified edX certificate (Read until the end for a tip on how to get a free certificate)

I finished the program a week ago, but I started it at the beginning of September (it means it took me 15 weeks to complete as a full-time student and part-time employee). Please plan accordingly because CS50AI is not a regular course with simple quizzes. It is very project-oriented. It expects you to go the extra mile.

Even though the course relates to CS50: Introduction to Computer Science, you would certainly need a better grasp of Python and Data Structures & Algorithms before you begin. It is crucial to know various Data Structures in Python. The course never teaches you to use Python dictionaries, sets, etc. The Python prerequisite for this course is very much real.

CS50AI should be completed only by those who have completed CS50x or equivalent; it does assume some background. Students also will likely benefit from some additional Python experience, either from CS50W or other comparable experience before taking this course. — CS50AI

Why is this particular course useful to your career as a beginner? Because it starts with the basics of Artificial Intelligence. You can’t run before you walk, so that’s why Week 0 of CS50AI is dedicated to basics algorithms like Breadth-First Search, Greedy Best-First Search, and Minimax.

You probably want to jump right to Machine Learning, which is on Week 4, but there are some necessary foundational topics before that, such as Propositional Logic (Week 1), Probability (Week 2), and Optimization (Week 3). You can rest assured that your journey will be smoother if you learn those topics before doing Machine Learning.

Now, let’s talk about the pros and cons.

CS50AI is comprehensive

As I have already mentioned, CS50AI starts with the necessary essential topics that will make you a more well-versed person in AI. For four weeks, before introducing the concepts of Machine Learning and Deep Learning, Brian Yu teaches the fundamental algorithms and concepts that serve as the foundation for later lectures. By making things clear in the beginning, the course lays a solid foundation for your future.

During those four weeks, you will accomplish the following projects:

  1. Applying the Breadth-First Search algorithm on the IMDb database to find the shortest connection between two actors.
  2. Using the Minimax decision rule to build an AI that never loses at Tic-Tac-Toe.
Demo Gif from my GitHub repo

3. Representing knowledge in propositional logic to build an AI that plays Minesweeper optimally.

Demo Gif from my GitHub repo

4. Simulating the basic version of Google Search Algorithm, PageRank.

5. And many other projects.

CS50AI is challenging

In the last three weeks of the course, Brian will introduce you to the cutting-edge technology dominating the field. Those are:

  1. Machine Learning, with the concepts of Supervised Learning, Unsupervised Learning, and Reinforcement Learning. You will actually get to build a reinforcement learning model based on Q-learning that will train AI to play the game of Nim. It was a brave step to include an assignment on reinforcement learning as it is not beginner-friendly, but believe me, it will be the first time you might think, “Is AI alive?”
  2. Deep Learning, with the concepts of Neural Networks, Computer Vision, and Convolutional Neural Networks. CS50AI will introduce you to the holy grail, TensorFlow. By the end of Week 5, you will build a Road Sign Recognition Project, using the provided data set, TensorFlow, and Open-CV. You will get hands-on experience with these awesome frameworks, such a treat.
  3. Natural Language Processing, with the concepts of Context-Free Grammar, Markov Models, and Naive Bayes. Ever thought about how Amazon Alexa understands your commands and responds to your questions like “How old is Bernie Sanders?” You will learn just that. One of the projects for Week 6 is building a system that responds to your questions using a popular library, NLTK.

CS50AI is everywhere

Just like its older brother, CS50, the course has communities everywhere. Discord, Quora, Twitter, Facebook, you name it. If you ever get stuck, any of those communities are there for you. If you have a problem installing a Python library, understanding the assignment, or debugging, there is always someone to ask.

This leads to the cons of taking the course.

CS50AI is long

You might have a natural question if there are so many challenging projects, how do they check my work? Unfortunately, due to the complexity of things, the projects are evaluated manually. Once you finish your project, you must record its functionality, upload it to YouTube, fill out another Google Form, and wait. How long does it take to receive a grade?

…we have your form and it will be graded within three weeks. — CS50AI

Yes, another human being has to check your code and see if it functions correctly, which is quite unusual for a MOOC. On the bright side, the person who checks your work provides some useful feedback if your submission did not pass.

Waiting three weeks for the grades to roll in is not an easy process. Especially if you cannot start the next week until you learn the grades for the current week, but it is worth waiting.

CS50AI is not as fun

If you have seen CS50 2019 with David Malan, you know how much fun it is. Watching Harvard students engaging in activities would bring the sense that you are a part of the classroom.

This is not the case with CS50AI. The pandemic had probably started by the time they recorded the videos, so it is only Brian you will see through the series. He does not make jokes (it is still interesting to watch, though).

Do it!

Now that I have mentioned the good and bad sides of the course, I can only say, “start taking it.” I do not think I have ever seen a course that would take you across all of the topics in AI and let you get hands-on experience in each. CS50AI will bring you to the basics of Machine Learning, Deep Learning, Natural Language Processing within the realms of trending Python and libraries like TensorFlow, Open-CV, scikit-learn, and NLTK.

Quick tip: You don’t have to pay for the edX Verified Certificate at all. CS50AI will provide you with one upon the completion of all their assignments. Their certificate looks like the one in the beginning, and you can share it on LinkedIn.

What to do after the course?

Take Andrew Ng’s Machine Learning course on Coursera if you are interested in Machine Learning.

Otherwise, take IBM’s Data Science on Coursera if you are interested in the more business-oriented use of data.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Damir Temir

Damir Temir


Data Science Aspirant w/ interest for writing. Student at the University of Illinois. Born and raised in Kazakhstan.