Welcome to MLST!

Machine Learning Street Talk is the top technical AI podcast on Spotify and YouTube. Thanks for finding us here!

WE ARE LOOKING FOR SPONSORS!

Email us to find out more - mlstreettalk at gmail.com

Would you like to make a one-off donation to MLST?

Just flying past? Are you looking for our fabulous AI YouTube channel or Podcast or public Discord channel, or perhaps our exclusive content on Patreon?


Hello! Tim Scarfe speaking here. You have reached our little corner of the internet. It took us a surprisingly long time to set up our online space for MLST. Better late than never though, right? We were so strapped for time early on that we simply didn’t have the will or inclination to do it.

We will use this cosy “ensconcement zone” for in-depth technical writing which will accompany the main audio podcast and YT channel. More importantly, we will use it as a collaborative thinking space to let content take shape creatively and collaboratively, before it goes out on the podcast or YT channel. We will publish articles “in the rough” so to speak, then continue to sculpt them in response to feedback from folks like you in the comments. Generally speaking the more iteration cycles and the more “intelligent entropy” we farm in, the better the result will be. Sound good?

Who Are We?

"Machine Learning Street Talk" co-hosts Tim Scarfe and Keith Duggar.

Dr. Tim Scarfe (left) started a successful software startup called Dot Net Solutions back in 2003, he has done stints at Microsoft as a Principal Engineer, Chief Data Scientist at bp, and founded several other tech startups and has a Ph.D in machine learning and a first-class degree in computer science.

Dr. Keith Duggar (right) who is often a co-interviewer on MLST — obtained his Ph.D from MIT and has done stints at IBM research, Wall Street, and Microsoft.

What is the ethos of MLST?

Intellectual curiosity.

We want to understand a broad range of views in the AI space, and this often involves getting our hands dirty — traversing deep and muddy philosophical trenches which undergird various ideas. We were covering many popular techniques years before they became mainstream, whether it was LLMs, active inference, neurosymbolic models or open-endedness.

Earlier on we were slightly more engineering-focussed and have drifted towards the theoretical since Yannic Kilcher’s departure but we may row back on that soon. Make no mistake though, we are engineers at heart. We often ground our analysis in computability theory, Physics, Bayesianism and cognitive philosophy. We often shift focus for periods of time, for example we have spent a lot of focus on Active Inference and Cognitive Science recently, before that — Linguistics.

We are in the extremely rare position that our salary does not depend on our technical opinion. We have the freedom to make content about anything which interests us. We love deep learning, but you won’t hear all the usual “tech bro” hype from us. We discovered the political “fault lines” of the AI community early on, it can be very hard to be opinionated and speak with a diverse group of people, but we will try our best to regardless.

The history of MLST and exciting news for 2024.

Machine Learning Street Talk grew out of a project which we started at Microsoft in 2018 called “Paper review calls”. We would run these fun weekly calls dissecting academic papers, often getting the actual researchers to present their work if from MSR. Sometimes - hundreds of Microsofties would join the calls! It was a rich community of some 6000 people, and Tim was recognised as the machine learning community champion of the year there in 2019. Admittedly, some of the recordings were cringeworthy but it was an incredible learning experience.

We launched the current incarnation of the project at the beginning of Covid in March 2021, teaming up with other fledgling YouTubers Yannic Kilcher and Connor Shorten. After about 60 episodes, Connor and Yannic departed to focus on their own personal projects — leaving Tim Scarfe running MLST with Keith Duggar on as a regular co-host. Keith and Tim co-founded a augmented reality company in 2021 which has taken up almost all of their time, so for much of 2023 MLST was in a bit of a holding pattern.

Doing MLST full-time in 2024

Tim Speaking. Here we are in 2024, and everything is about to change. Now that we have finished building the core technology for my startup company I co-founded with Keith — I have decided to step back and dedicate my full time and attention to MLST!

MLST was just a weekend project for me, but I have already built it up to be the number one technical AI podcast on Spotify — and we are about to hit 100K subscribers on YouTube, which is phenomenal!

This is only the beginning, the future will be exciting.

Be part of the MLST mission

This is a huge leap into the unknown for me. I would appreciate any support you can give me while I figure out how to make this project more viable commercially. I would love it if you supported our writing here by subscribing, but you may also do so over on our MLST patreon and you will also have access to all our subscriber-only writings. Also check out our business enquiries page for more options.

You can get in touch with Tim on LinkedIn or Discord.

Subscribe to Machine Learning Street Talk (MLST)

We are a media production company making video, audio and written content about machine learning, cognitive science and cognitive philosophy. Our podcast is the highest-rated technical AI podcast on Spotify.

People

I run the MLST podcast
Welcome! The team at MLST is inspired by academic research and each week we engage in dynamic discussion with pre-eminent figures in the AI field. We focus on AI, machine learning, cogsci and cogphil