Podcasts are a great way to learn about novel fields and tools, as well as keeping yourself updated with the fields that you care about.
I also believe that podcasts which are mainly centered around interviews are a great way to learn about the rockstars and superheroes of the AI world. You get a glimpse of how they think, what are they working on, and they solved a particular problem.
I would also argue that the content you get access to by listening to podcasts is very unique, and you cannot access them somewhere else.
In this post, I am not going into the details of why I think podcasts are great and Machine Learning learners and practitioners should listen to them.
Here are the podcasts that I highly recommend for ML learners and professionals-
from Weights and Biases, hosted by Lukas Biewald
It is a Machine Learning podcast that focuses on "Stories from machine learning experts solving real world problems".
Lukas hosts his podcasts in his own style. He asks really meaningful questions and is well-informed on the topics. The podcast episodes are very informative and engaging.
- Length per episode: usually 30-60 minutes
- Favorite episode so far: Peter Norvig, Google’s Director of Research – Singularity is in the eye of the beholder
from DeepMind, hosted by Hannah Fry
Hannah Fry is a superb host, as good as any professional podcast host. That sets this podcast apart from the others. Hanna Fry is a Mathematician and has had experience in hosting podcasts. She is passionate and knowledgeable about AI, and passionate about this podcast. This podcast likely has a team working behind it, and the sound effects and the music tells us so. They also make the experience absolutely entertaining.
This podcast focuses on works being done on DeepMind and hosts people working on those problems within DeepMind. The problems and solutions are described in plain words. The host also gives you a big picture view surrounding the problem, and what the future holds. This podcast is really well-structured, well-designed, and is really informative and edifying.
- Length per episode: 30 minutes
- Favorite episode so far: Episode 2: Go to Zero
hosted by Lex Fridman
Lex Fridman is an AI researcher working on autonomous vehicles, and human-robot interaction. He also teaches at MIT. The way he is unique as a podcast host is that he is very focused on long-term goals, and far-reaching implications of the works of the guests. He is interested in deep things such as consciousness and the idea of AGI. He asks deep, fundamental questions and lets us know about the guests' views on these questions.
The topics on his podcast transcend the AI universe. He has had guests like Roger Penrose, Richard Dawkins, and Noam Chomsky. A significant portion of his interviews are on AI. His interviews are generally long, and hence have much more scope to discuss topics in an in-depth, detailed manner. I really like that.
- Length per episode: 60-180 minutes
- Favorite episode so far: David Silver: AlphaGo, AlphaZero, and Deep Reinforcement Learning | Lex Fridman Podcast #86
hosted by Sanyam Bhutani
Sanyam Bhutani is a community man. Period. As a Data Scientist at h2o.ai, he writes blogs, hosts interviews, and hosts book-reading clubs and other clubs with Machine Learning Tokyo. He is also very active on the fast.ai Communities. And he never monetizes his content! When he is interviewing guests on his podcast, he is not asking questions for the interview, but for learning, too! I like his inquisitive, humble manner of asking questions.
This podcast hosts a lot of Kaggle Competition winners and Grandmasters. He talks about the guests' journey, career, and current research, projects, and interests. And, then there is Kaggle! He often interviews winners of particular competitions and asks about their thinking and approach to those particular competitions. These are really informative for people competing on Kaggle, and/or interested in developing full pipelines for any project. It is very interesting to see the diverse set of people from diverse backgrounds win Kaggle competitions, and talk about their careers and lives on this podcast.
- Length per episode: 30-90 minutes
- Favorite episode so far: Chris Deotte | Secrets to Becoming 4x Kaggle Grandmaster | Discussions and Notebooks
This is the only entry in the list which has multiple hosts. Yannic is famous for his paper-explaining videos, and his sunglasses. Dr. Scarfe and Keith also maintain an online presence and are involved in Deep Learning.
This podcast deliberately maintains a hacker, open-source sort of aura about them. That is evident from their choice of backgrounds, style of presentation, and many more things. They delve deep into research topics and interview the person(s) attached to them. The fact that there is more than one host adds value to the conversations, rather than being a liability. The listeners get to hear from multiple perspectives and get acquainted with multiple views. The guests are also much more likely to shed light on different aspects of the research.
- Length per episode: 90-120 minutes
- Favorite episode so far: #040 - Adversarial Examples
The +1. Talks at Google, Microsoft Research, and TED/TEDx
Of course, you have noticed that "+1" in the title. I am including these resources together as one, because, all of the content present at the first two resources are not focused on ML or AI. And although TED has a few talks based on this topic and broader topics, it has some really good talks on ML, Data Science, and AI.
If you visit the Microsoft Research YouTube page, you will see many playlists that will kindle your excitement. For example, the AI Distinguished Lecture Series contains many thought-provoking talks in AI. The same goes with Talks at Google.
TED has quite a few talks on Data Science, AI, and its impact. And TED has been covering Data Analysis, AI, etc. years before these fields blew up. I remember listening to a talk- Making data mean more through storytelling | Ben Wellington which was given back in 2015 that highlights how creative, yet simple Data Analysis can reveal many important insights, and how it can make lives better by influencing government policies. There's another talk that I will mention- Daphne Koller: What we're learning from online education. The co-founder of Coursera gives us many deep insights about online education and how data are at the core and crux of it. This is yet another example of data transforming something fundamental like education, and impacting the lives of millions positively.
- Length per episode: mixed
Podcasts about and around Machine Learning have seen a boom in recent times. It is important not to get overwhelmed in this overflow. I have listened to many podcasts about ML/Data Science and decided on these as the bests.
If you have any suggestions or questions, please let me know in the comments.